title

Artificial Intelligence Full Course | Artificial Intelligence Tutorial for Beginners | Edureka

description

๐ฅ Machine Learning Engineer Masters Program (Use Code "๐๐๐๐๐๐๐๐๐"): https://www.edureka.co/masters-program/machine-learning-engineer-training
This Edureka video on "Artificial Intelligence Full Course" will provide you with a comprehensive and detailed knowledge of Artificial Intelligence concepts with hands-on examples. The following topics are covered in this Artificial Intelligence Full Course:
00:00 Introduction to Artificial Intelligence Course
02:27 History Of AI
06:45 Demand For AI
08:46 What Is Artificial Intelligence?
09:50 AI Applications
16:49 Types Of AI
20:24 Programming Languages For AI
27:12 Introduction To Machine Learning
28:08 Need For Machine Learning
31:48 What Is Machine Learning?
34:13 Machine Learning Definitions
37:26 Machine Learning Process
49:13 Types Of Machine Learning
49:21 Supervised Learning
52:00 Unsupervised Learning
53:44 Reinforcement Learning
55:29 Supervised vs Unsupervised vs Reinforcement Learning
58:23 Types Of Problems Solved Using Machine Learning
1:04:49 Supervised Learning Algorithms
1:05:17 Linear Regression
1:11:20 Linear Regression Demo
1:26:36 Logistic Regression
1:35:36 Decision Tree
1:55:18 Random Forest
2:07:31 Naive Bayes
2:14:37 K Nearest Neighbour (KNN)
2:20:31 Support Vector Machine (SVM)
2:26:40 Demo (Classification Algorithms)
2:42:36 Unsupervised Learning Algorithms
2:42:45 K-means Clustering
2:50:49 Demo (Unsupervised Learning)
2:56:40 Reinforcement Learning
3:24:36 Demo (Reinforcement Learning)
3:31:41 AI vs Machine Learning vs Deep Learning
3:33:08 Limitations Of Machine Learning
3:36:32 Introduction To Deep Learning
3:38:36 How Deep Learning Works?
3:40:48 What Is Deep Learning?
3:41:50 Deep Learning Use Case
3:43:14 Single Layer Perceptron
3:50:56 Multi Layer Perceptron (ANN)
3:52:55 Backpropagation
3:54:39 Training A Neural Network
4:01:02 Limitations Of Feed Forward Network
4:03:18 Recurrent Neural Networks
4:05:36 Convolutional Neural Networks
4:09:00 Demo (Deep Learning)
4:29:02 Natural Language Processing
4:30:53 What Is Text Mining?
4:32:43 What Is NLP?
4:33:26 Applications Of NLP
4:35:53 Terminologies In NLP
4:41:19 NLP Demo
4:47:21 Machine Learning Masters Program
Python Full Course: https://www.youtube.com/watch?v=WGJJIrtnfpk
Statistics and Probability Tutorial: https://www.youtube.com/watch?v=XcLO4f1i4Yo
๐ด ๐๐๐ฎ๐ซ๐๐ค๐ ๐๐ฒ๐ญ๐ก๐จ๐ง ๐๐ซ๐๐ข๐ง๐ข๐ง๐ ๐ฌ
๐ต Python Programming Certification: http://bit.ly/37rEsnA
๐ต Python Certification Training for Data Science: http://bit.ly/2Gj6fux
๐ด. ๐๐๐ฎ๐ซ๐๐ค๐ ๐๐๐ฌ๐ญ๐๐ซ๐ฌ ๐๐ซ๐จ๐ ๐ซ๐๐ฆ
๐ต Data Scientist Masters Program: http://bit.ly/2t1snGM
๐ต Machine Learning Engineer Masters Program: https://bit.ly/3Hi1sXN
๐ด ๐๐๐ฎ๐ซ๐๐ค๐ ๐๐ง๐ข๐ฏ๐๐ซ๐ฌ๐ข๐ญ๐ฒ ๐๐ซ๐จ๐ ๐ซ๐๐ฆ
๐ต Advanced Certificate Program in Data Science with E&ICT Academy, IIT Guwahati: http://bit.ly/3V7ffrh
๐ต University of Cambridge Online Certifications: https://bit.ly/3RSNTXi
๐ข๐ข ๐๐จ๐ฉ ๐๐ ๐๐ซ๐๐ง๐๐ข๐ง๐ ๐๐๐๐ก๐ง๐จ๐ฅ๐จ๐ ๐ข๐๐ฌ ๐ญ๐จ ๐๐๐๐ซ๐ง ๐ข๐ง ๐๐๐๐ ๐๐๐ซ๐ข๐๐ฌ ๐ข๐ข
โฉ NEW Top 10 Technologies To Learn In 2024 - https://www.youtube.com/watch?v=vaLXPv0ewHU
Do subscribe to our channel and hit the bell icon to never miss an update from us in the future: https://goo.gl/6ohpTV
Check out the entire Machine Learning Playlist: https://bit.ly/2NG9tK4
Check out the entire Machine Learning Blog list: https://bit.ly/2V2MnDW
#edureka #artificialIntelligence #artificialIntelligenceTutorial #artificialIntelligenceFullCourse #artificialIntelligenceEngineer
๐๐๐๐ฅ๐๐ ๐ซ๐๐ฆ: https://t.me/edurekaupdates
๐๐๐ฐ๐ข๐ญ๐ญ๐๐ซ: https://twitter.com/edurekain
๐๐๐ข๐ง๐ค๐๐๐๐ง: https://www.linkedin.com/company/edureka
๐๐๐ง๐ฌ๐ญ๐๐ ๐ซ๐๐ฆ: https://www.instagram.com/edureka_learning/
๐๐
๐๐๐๐๐จ๐จ๐ค: https://www.facebook.com/edurekaIN/
๐๐๐ฅ๐ข๐๐๐๐ก๐๐ซ๐: https://www.slideshare.net/EdurekaIN
๐๐๐๐ฌ๐ญ๐๐จ๐ฑ: https://castbox.fm/networks/505?country=IN
๐๐๐๐๐ญ๐ฎ๐ฉ: https://www.meetup.com/edureka/
๐๐๐จ๐ฆ๐ฆ๐ฎ๐ง๐ข๐ญ๐ฒ: https://www.edureka.co/community/
-------------------------------------
About the Masters Program
Edurekaโs Machine Learning Engineer Masters Program makes you proficient in techniques like Supervised Learning, Unsupervised Learning and Natural Language Processing. It includes training on the latest advancements and technical approaches in Artificial Intelligence & Machine Learning such as Deep Learning, Graphical Models and Reinforcement Learning.
The Master's Program Covers Topics LIke:
Python Programming
PySpark
HDFS
Spark SQL
Machine Learning Techniques and Artificial Intelligence Types
Tokenization
Named Entity Recognition
Lemmatization
Supervised Algorithms
Unsupervised Algorithms
Tensor Flow
Deep learning
Keras
Neural Networks
Bayesian and Markovโs Models
Inference
Decision Making
Bandit Algorithms
Bellman Equation
Policy Gradient Methods.
-------------
Please write back to us at sales@edureka.in or call us at IND: 9606058406 / US: +18338555775 (toll-free) for more information

detail

{'title': 'Artificial Intelligence Full Course | Artificial Intelligence Tutorial for Beginners | Edureka', 'heatmap': [{'end': 704.199, 'start': 524.309, 'weight': 0.816}, {'end': 2461.395, 'start': 1929.631, 'weight': 0.928}, {'end': 4043.087, 'start': 3513.18, 'weight': 0.854}, {'end': 6333.297, 'start': 6147.665, 'weight': 0.773}, {'end': 7214.324, 'start': 7028.58, 'weight': 0.754}, {'end': 13181.362, 'start': 12649.068, 'weight': 1}, {'end': 14063.358, 'start': 13876.823, 'weight': 0.702}, {'end': 14585.826, 'start': 14405.707, 'weight': 0.735}], 'summary': 'The full course on artificial intelligence covers the history, applications, and basics of ai, machine learning, deep learning, nlp using python, and real-world applications of ai in various sectors. it discusses machine learning importance in handling vast data, types and problems of machine learning, regression and classification algorithms, random forest, naive bias, k-nearest neighbor, reinforcement learning, deep learning, and nlp with specific examples and supervised learning methods, emphasizing the significance of labeled data in supervised learning.', 'chapters': [{'end': 134.498, 'segs': [{'end': 134.498, 'src': 'embed', 'start': 86.62, 'weight': 0, 'content': [{'end': 93.085, 'text': "We'll run a couple of demos wherein we'll see how machine learning algorithms are used to solve real-world problems after that.", 'start': 86.62, 'duration': 6.465}, {'end': 97.169, 'text': "We'll discuss the limitations of machine learning and why deep learning is needed.", 'start': 93.125, 'duration': 4.044}, {'end': 99.864, 'text': "I'll introduce you to the deep learning concepts.", 'start': 97.722, 'duration': 2.142}, {'end': 104.287, 'text': 'What are neurons, perceptrons, multiple layer perceptrons and so on?', 'start': 100.004, 'duration': 4.283}, {'end': 109.891, 'text': "we'll discuss the different types of neural networks and we'll also look at what exactly back propagation is.", 'start': 104.287, 'duration': 5.604}, {'end': 110.491, 'text': 'apart from this,', 'start': 109.891, 'duration': 0.6}, {'end': 117.937, 'text': "We'll be running a demo to understand deep learning in more depth and finally we'll move on to the next module which is natural language processing.", 'start': 110.531, 'duration': 7.406}, {'end': 119.799, 'text': 'Under natural language processing.', 'start': 118.397, 'duration': 1.402}, {'end': 124.085, 'text': "We'll try to understand what is text mining the difference between text mining and NLP.", 'start': 119.839, 'duration': 4.246}, {'end': 131.754, 'text': "What are the different terminologies in NLP and we'll end the session by looking at a practical implementation of NLP using python.", 'start': 124.465, 'duration': 7.289}, {'end': 134.498, 'text': "All right, so guys there's a lot to cover in today's session.", 'start': 132.075, 'duration': 2.423}], 'summary': 'Demos on machine learning and deep learning, followed by introduction to natural language processing with a practical implementation using python.', 'duration': 47.878, 'max_score': 86.62, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk86620.jpg'}], 'start': 11.192, 'title': 'Ai full course overview', 'summary': 'Covers the history, applications, and basics of artificial intelligence, including machine learning and deep learning concepts, with a practical implementation of nlp using python, in a comprehensive session on artificial intelligence full course.', 'chapters': [{'end': 134.498, 'start': 11.192, 'title': 'Ai full course overview', 'summary': 'Covers the history, applications, and basics of artificial intelligence, including machine learning and deep learning concepts, with a practical implementation of nlp using python, in a comprehensive session on artificial intelligence full course.', 'duration': 123.306, 'highlights': ['The session covers the history, applications, and basics of artificial intelligence, including machine learning and deep learning concepts, with a practical implementation of NLP using Python.', 'The different types of machine learning, algorithms involved, and demos showcasing real-world problem solving using machine learning will be discussed.', 'Understanding the different types of neural networks, back propagation, and a deep learning demo will be covered in the session.', 'An overview of natural language processing, including text mining, NLP terminologies, and a practical NLP implementation using Python, will be presented.']}], 'duration': 123.306, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11192.jpg', 'highlights': ['The session covers the history, applications, and basics of artificial intelligence, including machine learning and deep learning concepts, with a practical implementation of NLP using Python.', 'Understanding the different types of neural networks, back propagation, and a deep learning demo will be covered in the session.', 'The different types of machine learning, algorithms involved, and demos showcasing real-world problem solving using machine learning will be discussed.', 'An overview of natural language processing, including text mining, NLP terminologies, and a practical NLP implementation using Python, will be presented.']}, {'end': 945.483, 'segs': [{'end': 181.163, 'src': 'embed', 'start': 155.828, 'weight': 2, 'content': [{'end': 163.011, 'text': 'under Greek mythology, the concept of machines and mechanical men were well thought of, so an example of this is Talos.', 'start': 155.828, 'duration': 7.183}, {'end': 170.894, 'text': "I don't know how many of you have heard of this Talos was a giant animated bronze Warrior was programmed to guard the island of Crete.", 'start': 163.351, 'duration': 7.543}, {'end': 172.815, 'text': 'Now, these are just ideas.', 'start': 171.534, 'duration': 1.281}, {'end': 178.517, 'text': 'Nobody knows if this was actually implemented but machine learning and AI were thought of long ago.', 'start': 172.955, 'duration': 5.562}, {'end': 181.163, 'text': "Now let's get back to the 19th century.", 'start': 179.122, 'duration': 2.041}], 'summary': 'Greek mythology includes the concept of a giant animated bronze warrior, talos, programmed to guard crete, reflecting early ideas of machines and ai.', 'duration': 25.335, 'max_score': 155.828, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk155828.jpg'}, {'end': 387.686, 'src': 'embed', 'start': 355.269, 'weight': 1, 'content': [{'end': 357.871, 'text': 'Champions Brad Ritter and Ken Jennings.', 'start': 355.269, 'duration': 2.602}, {'end': 363.396, 'text': 'So guys this was how AI evolved it started off as a hypothetical situation right now.', 'start': 358.431, 'duration': 4.965}, {'end': 366.54, 'text': "It's the most important technology in today's world, right?", 'start': 363.496, 'duration': 3.044}, {'end': 371.545, 'text': 'If you look around everywhere, everything around us is run through AI, deep learning or machine learning.', 'start': 366.6, 'duration': 4.945}, {'end': 378.552, 'text': 'So since the emergence of AI in the 1950s, we have actually seen an exponential growth in his potential.', 'start': 372.105, 'duration': 6.447}, {'end': 387.686, 'text': 'So AI covers domains such as machine learning deep learning neural networks natural language processing knowledge base expert systems and so on.', 'start': 379.142, 'duration': 8.544}], 'summary': 'Ai has evolved since the 1950s, now vital in many domains.', 'duration': 32.417, 'max_score': 355.269, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk355269.jpg'}, {'end': 704.199, 'src': 'heatmap', 'start': 524.309, 'weight': 0.816, 'content': [{'end': 528.331, 'text': "So let's move on and understand what exactly artificial intelligence is.", 'start': 524.309, 'duration': 4.022}, {'end': 536.135, 'text': 'The term artificial intelligence was first coined in the year 1956 by John McCarthy at the Dartmouth Conference.', 'start': 529.051, 'duration': 7.084}, {'end': 538.056, 'text': 'I already mentioned this before.', 'start': 536.795, 'duration': 1.261}, {'end': 541.718, 'text': 'It was the birth of AI in 1956.', 'start': 538.576, 'duration': 3.142}, {'end': 550.042, 'text': 'Now how did he define artificial intelligence? John McCarthy defined AI as the science and engineering of making intelligent machines.', 'start': 541.718, 'duration': 8.324}, {'end': 551.849, 'text': 'In other words,', 'start': 550.868, 'duration': 0.981}, {'end': 560.219, 'text': 'artificial intelligence is the theory and development of computer systems able to perform tasks that normally require human intelligence,', 'start': 551.849, 'duration': 8.37}, {'end': 566.065, 'text': 'such as visual perception, speech recognition, decision-making and translation between languages.', 'start': 560.219, 'duration': 5.846}, {'end': 572.032, 'text': 'So guys in a sense AI is a technique of getting machines to work and behave like humans.', 'start': 566.646, 'duration': 5.386}, {'end': 573.712, 'text': 'In the recent past,', 'start': 572.632, 'duration': 1.08}, {'end': 581.154, 'text': 'artificial intelligence has been able to accomplish this by creating machines and robots that have been used in wide range of fields,', 'start': 573.712, 'duration': 7.442}, {'end': 586.215, 'text': 'including healthcare, robotics, marketing, business analytics and many more.', 'start': 581.154, 'duration': 5.061}, {'end': 586.875, 'text': 'with this in mind,', 'start': 586.215, 'duration': 0.66}, {'end': 593.617, 'text': "Let's discuss a couple of real-world applications of AI so that you understand how important artificial intelligence is in today's world.", 'start': 586.895, 'duration': 6.722}, {'end': 600.042, 'text': 'Now, one of the most famous applications of artificial intelligence is the Google predictive search engine.', 'start': 594.301, 'duration': 5.741}, {'end': 607.403, 'text': 'when you begin typing a search term and Google makes recommendations for you to choose from, that is artificial intelligence in action.', 'start': 600.042, 'duration': 7.361}, {'end': 615.265, 'text': 'So predictive searches are based on data that Google collects about you, such as your browser history, your location,', 'start': 607.983, 'duration': 7.282}, {'end': 617.165, 'text': 'your age and other personal details.', 'start': 615.265, 'duration': 1.9}, {'end': 622.946, 'text': 'So by using artificial intelligence Google attempts to guess what you might be trying to find.', 'start': 617.685, 'duration': 5.261}, {'end': 628.479, 'text': "Now behind this there's a lot of natural language processing deep learning and machine learning involved.", 'start': 623.712, 'duration': 4.767}, {'end': 632.045, 'text': "We'll be discussing all of those concepts in the further slides, right?", 'start': 628.92, 'duration': 3.125}, {'end': 638.415, 'text': "It's not very simple to create a search engine, but the logic behind Google search engine is artificial intelligence.", 'start': 632.065, 'duration': 6.35}, {'end': 641.44, 'text': 'Moving on in the finance sector.', 'start': 639.179, 'duration': 2.261}, {'end': 646.463, 'text': "JP Morgan's Chase contract intelligence platform uses machine learning,", 'start': 641.44, 'duration': 5.023}, {'end': 651.206, 'text': 'artificial intelligence and image recognition software to analyze legal documents.', 'start': 646.463, 'duration': 4.743}, {'end': 658.49, 'text': 'Now, let me tell you that a manually reviewing around 12, 000 agreements took over 36, 000 hours.', 'start': 651.706, 'duration': 6.784}, {'end': 659.55, 'text': "That's a lot of time.", 'start': 658.73, 'duration': 0.82}, {'end': 665.874, 'text': 'But as soon as this task was replaced by a AI machine, it was able to do this in a matter of seconds.', 'start': 659.971, 'duration': 5.903}, {'end': 670.655, 'text': "So that's the difference between artificial intelligence and manual or human work.", 'start': 666.494, 'duration': 4.161}, {'end': 678.638, 'text': 'Even though AI cannot think and reason like humans, but their computational power is very strong compared to humans.', 'start': 671.316, 'duration': 7.322}, {'end': 683.619, 'text': 'right because of machine learning algorithms, deep learning concepts and natural language processing,', 'start': 678.638, 'duration': 4.981}, {'end': 689.861, 'text': 'AI has reached a stage wherein it can compute the most complex to complex problems in a matter of seconds.', 'start': 683.619, 'duration': 6.242}, {'end': 696.043, 'text': 'coming to health care, IBM is one of the Pioneers that has developed AI software specifically for medicine.', 'start': 689.861, 'duration': 6.182}, {'end': 704.199, 'text': 'Let me tell you that more than 230 healthcare organizations use IBM AI technology, which is basically IBM Watson.', 'start': 696.79, 'duration': 7.409}], 'summary': "Artificial intelligence, coined in 1956, enables machines to perform human-like tasks. ai powers google's predictive search and jp morgan's contract analysis, saving 36,000 hours of manual work.", 'duration': 179.89, 'max_score': 524.309, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk524309.jpg'}, {'end': 600.042, 'src': 'embed', 'start': 573.712, 'weight': 0, 'content': [{'end': 581.154, 'text': 'artificial intelligence has been able to accomplish this by creating machines and robots that have been used in wide range of fields,', 'start': 573.712, 'duration': 7.442}, {'end': 586.215, 'text': 'including healthcare, robotics, marketing, business analytics and many more.', 'start': 581.154, 'duration': 5.061}, {'end': 586.875, 'text': 'with this in mind,', 'start': 586.215, 'duration': 0.66}, {'end': 593.617, 'text': "Let's discuss a couple of real-world applications of AI so that you understand how important artificial intelligence is in today's world.", 'start': 586.895, 'duration': 6.722}, {'end': 600.042, 'text': 'Now, one of the most famous applications of artificial intelligence is the Google predictive search engine.', 'start': 594.301, 'duration': 5.741}], 'summary': 'Ai has wide applications including healthcare, robotics, marketing; google predictive search is a notable example.', 'duration': 26.33, 'max_score': 573.712, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk573712.jpg'}], 'start': 135.018, 'title': 'Evolution of ai and its impact', 'summary': "Traces the history of artificial intelligence from classical ages to recent years, highlighting key milestones, including the turing test and real-world applications in various sectors such as google's predictive search, finance, healthcare, social media, virtual assistants, and self-driving cars.", 'chapters': [{'end': 178.517, 'start': 135.018, 'title': 'History of artificial intelligence', 'summary': 'Discusses the history of artificial intelligence, tracing back to the classical ages, with a mention of talos, a giant animated bronze warrior programmed to guard the island of crete, reflecting the early concepts of machine learning and ai.', 'duration': 43.499, 'highlights': ['The concept of artificial intelligence dates back to the classical ages, as seen in Greek mythology with the example of Talos, a giant animated bronze Warrior programmed to guard the island of Crete.', 'Machine learning and AI were conceptualized long ago, evident in the ideas of machines and mechanical men in Greek mythology.', 'Encourages audience to subscribe to the YouTube channel for updates on recent Technologies and training sessions.']}, {'end': 945.483, 'start': 179.122, 'title': 'Evolution of ai and its impact', 'summary': "Traces the development of artificial intelligence from the 1950s, highlighting key milestones such as the creation of the turing test, early game ai, and significant achievements in recent years. it explains the surge in ai's importance, attributing it to factors such as increased computational power, growing data availability, advancements in algorithms, and heavy investments from various sectors. real-world applications of ai in google's predictive search, finance, healthcare, social media, virtual assistants, and self-driving cars are also discussed.", 'duration': 766.361, 'highlights': ["AI's Evolution from 1950s to Present", "Importance of AI in Today's World", 'Real-World Applications of AI']}], 'duration': 810.465, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk135018.jpg', 'highlights': ['Real-World Applications of AI', "AI's Evolution from 1950s to Present", 'The concept of artificial intelligence dates back to the classical ages, as seen in Greek mythology with the example of Talos, a giant animated bronze Warrior programmed to guard the island of Crete.', 'Machine learning and AI were conceptualized long ago, evident in the ideas of machines and mechanical men in Greek mythology.']}, {'end': 3188.475, 'segs': [{'end': 995.695, 'src': 'embed', 'start': 964.148, 'weight': 1, 'content': [{'end': 971.137, 'text': 'For example, we have primary section, social section, and all of that Gmail has a separate section called the spam meals also.', 'start': 964.148, 'duration': 6.989}, {'end': 981.05, 'text': 'So what Gmail does is it makes use of concepts of artificial intelligence and machine learning algorithms to classify emails as spam and non-spam.', 'start': 971.747, 'duration': 9.303}, {'end': 985.972, 'text': 'many a time, certain words or phrases are frequently used in spam emails.', 'start': 981.05, 'duration': 4.922}, {'end': 995.695, 'text': 'If you notice your spam emails, they have words like lottery on full refund all of this denotes that the email is more likely to be a spam one.', 'start': 986.532, 'duration': 9.163}], 'summary': "Gmail uses ai to classify spam and non-spam emails based on frequent spam words like 'lottery' and 'full refund'.", 'duration': 31.547, 'max_score': 964.148, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk964148.jpg'}, {'end': 1320.204, 'src': 'embed', 'start': 1292.504, 'weight': 0, 'content': [{'end': 1300.791, 'text': 'So python is considered the best choice for artificial intelligence with python stands are which is a statistical programming language.', 'start': 1292.504, 'duration': 8.287}, {'end': 1307.957, 'text': 'Now R is one of the most effective language and environment for analyzing and manipulating the data for statistical purpose.', 'start': 1301.291, 'duration': 6.666}, {'end': 1310.679, 'text': 'It is a statistical programming language.', 'start': 1308.677, 'duration': 2.002}, {'end': 1320.204, 'text': 'So using R we can easily produce well-designed publication quality plots including mathematical symbols and formulae wherever needed.', 'start': 1311.159, 'duration': 9.045}], 'summary': 'Python is best for ai, r for statistical data analysis and plot production.', 'duration': 27.7, 'max_score': 1292.504, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk1292504.jpg'}, {'end': 1762.568, 'src': 'embed', 'start': 1739.688, 'weight': 2, 'content': [{'end': 1752.079, 'text': 'So top-tier companies like Netflix and Amazon build such machine learning models by using tons of data in order to identify any profitable opportunity and avoid any unwanted risk.', 'start': 1739.688, 'duration': 12.391}, {'end': 1759.385, 'text': "So, guys, one thing you'll need to know is that the most important thing for artificial intelligence is data for artificial intelligence,", 'start': 1752.559, 'duration': 6.826}, {'end': 1761.447, 'text': "or whether it's machine learning or deep learning.", 'start': 1759.385, 'duration': 2.062}, {'end': 1762.568, 'text': "It's always data.", 'start': 1761.907, 'duration': 0.661}], 'summary': 'Top-tier companies use tons of data for machine learning models to identify profitable opportunities and mitigate risks.', 'duration': 22.88, 'max_score': 1739.688, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk1739688.jpg'}, {'end': 2461.395, 'src': 'heatmap', 'start': 1929.631, 'weight': 0.928, 'content': [{'end': 1933.512, 'text': 'because most of the technologies today are based on the concept of machine learning.', 'start': 1929.631, 'duration': 3.881}, {'end': 1939.587, 'text': 'Most of the AI Technologies itself are based on the concept of machine learning and deep learning.', 'start': 1934.241, 'duration': 5.346}, {'end': 1942.27, 'text': "don't get confused about machine learning and deep learning.", 'start': 1939.587, 'duration': 2.683}, {'end': 1948.497, 'text': "We'll discuss about deep learning in the further slides where we'll also see the difference between AI machine learning and deep learning.", 'start': 1942.63, 'duration': 5.867}, {'end': 1951.78, 'text': 'So coming back to what exactly machine learning is.', 'start': 1949.179, 'duration': 2.601}, {'end': 1957.562, 'text': "If you browse through the internet, you'll find a lot of definitions about what exactly machine learning is.", 'start': 1951.8, 'duration': 5.762}, {'end': 1970.848, 'text': 'One of the definitions I found was a computer program is said to learn from experience E with respect to some class of task T and performance measure P if its performance at tasks in T,', 'start': 1958.063, 'duration': 12.785}, {'end': 1973.969, 'text': 'as measured by P, improves with experience E.', 'start': 1970.848, 'duration': 3.121}, {'end': 1975.75, 'text': "That's very confusing.", 'start': 1974.549, 'duration': 1.201}, {'end': 1979.352, 'text': 'So let me just narrow it down to you in simple terms.', 'start': 1975.83, 'duration': 3.522}, {'end': 1991.18, 'text': 'machine learning is a subset of artificial intelligence which provides machines the ability to learn automatically and improve from experience without being explicitly programmed to do so.', 'start': 1979.352, 'duration': 11.828}, {'end': 1997.987, 'text': 'In the sense it is the practice of getting machines to solve problems by gaining the ability to think.', 'start': 1991.821, 'duration': 6.166}, {'end': 1999.609, 'text': 'but now you might be thinking.', 'start': 1997.987, 'duration': 1.622}, {'end': 2002.651, 'text': 'how can a machine think or make decisions?', 'start': 1999.609, 'duration': 3.042}, {'end': 2004.313, 'text': 'now machines are very similar to humans.', 'start': 2002.651, 'duration': 1.662}, {'end': 2013.582, 'text': 'Okay, if you feed a machine a good amount of data, it will learn how to interpret, process and analyze this data by using machine learning algorithms,', 'start': 2004.473, 'duration': 9.109}, {'end': 2015.644, 'text': 'and it will help you solve real-world problems.', 'start': 2013.582, 'duration': 2.062}, {'end': 2019.83, 'text': 'So what happens here is a lot of data is fed to the machine.', 'start': 2016.224, 'duration': 3.606}, {'end': 2031.208, 'text': 'The machine will train on this data and it will build a predictive model with the help of machine learning algorithms in order to predict some outcome or in order to find some solution to a problem.', 'start': 2020.271, 'duration': 10.937}, {'end': 2033.042, 'text': 'So it involves data.', 'start': 2031.781, 'duration': 1.261}, {'end': 2041.71, 'text': "You're going to train the machine and build a model by using machine learning algorithms in order to predict some outcome or to find a solution to a problem.", 'start': 2033.302, 'duration': 8.408}, {'end': 2045.954, 'text': 'So that is a simple way of understanding what exactly machine learning is.', 'start': 2042.21, 'duration': 3.744}, {'end': 2049.116, 'text': "I'll be going into more depth about machine learning.", 'start': 2046.574, 'duration': 2.542}, {'end': 2051.859, 'text': "So don't worry if you haven't understood anything as of now.", 'start': 2049.237, 'duration': 2.622}, {'end': 2056.87, 'text': "Now let's discuss a couple of terms which are frequently used in machine learning right?", 'start': 2052.824, 'duration': 4.046}, {'end': 2062.976, 'text': 'So the first definition that we come across very often is an algorithm right?', 'start': 2057.291, 'duration': 5.685}, {'end': 2064.079, 'text': 'So, basically,', 'start': 2063.398, 'duration': 0.681}, {'end': 2073.17, 'text': 'a machine learning algorithm is a set of rules and statistical techniques that is used to learn patterns from data and draw significant information from it.', 'start': 2064.079, 'duration': 9.091}, {'end': 2078.734, 'text': 'Okay, so guys the logic behind a machine learning model is basically the machine learning algorithm.', 'start': 2073.71, 'duration': 5.024}, {'end': 2085.339, 'text': 'Okay, an example of a machine learning algorithm is linear regression or decision tree or a random forest.', 'start': 2079.215, 'duration': 6.124}, {'end': 2090.864, 'text': 'All of these are machine learning algorithms which define the logic behind a machine learning model.', 'start': 2085.639, 'duration': 5.225}, {'end': 2096.829, 'text': 'Now what is a machine learning model? A model is actually the main component of a machine learning process.', 'start': 2091.264, 'duration': 5.565}, {'end': 2100.772, 'text': 'Okay, so a model is trained by using the machine learning algorithm.', 'start': 2097.249, 'duration': 3.523}, {'end': 2112.113, 'text': 'Now, the difference between an algorithm and a model is that an algorithm maps all the decisions that a model is supposed to take based on the given input in order to get the correct output.', 'start': 2101.313, 'duration': 10.8}, {'end': 2122.2, 'text': 'So the model will use the machine learning algorithm in order to draw useful insights from the input and give you an outcome that is very precise.', 'start': 2113.354, 'duration': 8.846}, {'end': 2123.54, 'text': "That's the machine learning model.", 'start': 2122.24, 'duration': 1.3}, {'end': 2126.662, 'text': 'The next definition we have is predictor variable.', 'start': 2124.021, 'duration': 2.641}, {'end': 2132.586, 'text': 'Now a predictor variable is any feature of the data that can be used to predict the output.', 'start': 2127.103, 'duration': 5.483}, {'end': 2137.491, 'text': 'Okay, let me give you an example to make you understand what a predictor variable is.', 'start': 2133.246, 'duration': 4.245}, {'end': 2143.076, 'text': "Let's say you're trying to predict the height of a person depending on his weight, right?", 'start': 2137.951, 'duration': 5.125}, {'end': 2150.744, 'text': "So here your predictor variable becomes your weight, because you're using the weight of a person to predict the person's height.", 'start': 2143.297, 'duration': 7.447}, {'end': 2153.387, 'text': 'So your predictor variable becomes your weight.', 'start': 2151.105, 'duration': 2.282}, {'end': 2156.113, 'text': 'The next definition is response variable.', 'start': 2154.071, 'duration': 2.042}, {'end': 2159.696, 'text': 'Now in the same example, height would be the response variable.', 'start': 2156.453, 'duration': 3.243}, {'end': 2165.2, 'text': 'Response variable is also known as the target variable or the output variable.', 'start': 2160.056, 'duration': 5.144}, {'end': 2169.083, 'text': "This is the variable that you're trying to predict by using the predictor variables.", 'start': 2165.32, 'duration': 3.763}, {'end': 2175.929, 'text': 'So a response variable is a feature or the output variable that needs to be predicted by using the predictor variables.', 'start': 2169.604, 'duration': 6.325}, {'end': 2179.252, 'text': 'Next we have something known as training data.', 'start': 2177.29, 'duration': 1.962}, {'end': 2185.057, 'text': "Now training and testing data are terminologies that you'll come across very often in a machine learning process.", 'start': 2179.612, 'duration': 5.445}, {'end': 2190.922, 'text': 'So training data is basically the data that is used to create the machine learning model.', 'start': 2185.478, 'duration': 5.444}, {'end': 2196.868, 'text': "So basically in a machine learning process, when you feed data to the machine, it'll be divided into two parts.", 'start': 2191.223, 'duration': 5.645}, {'end': 2201.692, 'text': 'All right, so splitting the data into two parts is also known as data splicing.', 'start': 2197.308, 'duration': 4.384}, {'end': 2204.995, 'text': "So you'll take your input data, you'll divide it into two sections.", 'start': 2201.992, 'duration': 3.003}, {'end': 2209.318, 'text': "One you'll call the training data and the other you'll call the testing data.", 'start': 2205.475, 'duration': 3.843}, {'end': 2212.081, 'text': 'So then you have something known as the testing data.', 'start': 2209.819, 'duration': 2.262}, {'end': 2216.685, 'text': 'The training data is basically used to create the machine learning model.', 'start': 2212.481, 'duration': 4.204}, {'end': 2222.59, 'text': 'The training data helps the model to identify key trends and patterns which are essential to predict the output.', 'start': 2216.765, 'duration': 5.825}, {'end': 2231.615, 'text': 'Now the testing data is after the model is trained, it must be tested in order to evaluate how accurately it can predict an outcome.', 'start': 2223.13, 'duration': 8.485}, {'end': 2233.797, 'text': 'Now this is done by using the testing data.', 'start': 2231.775, 'duration': 2.022}, {'end': 2236.758, 'text': 'So basically the training data is used to train the model.', 'start': 2234.257, 'duration': 2.501}, {'end': 2240.2, 'text': 'The testing data is used to test the efficiency of the model.', 'start': 2236.938, 'duration': 3.262}, {'end': 2245.924, 'text': "Now let's move on and look at our next topic which is machine learning process.", 'start': 2241.221, 'duration': 4.703}, {'end': 2248.57, 'text': 'So what is the machine learning process?', 'start': 2246.528, 'duration': 2.042}, {'end': 2256.099, 'text': 'Now, the machine learning process involves building a predictive model that can be used to find a solution for a problem statement.', 'start': 2249.051, 'duration': 7.048}, {'end': 2261.926, 'text': 'Now in order to solve any problem in machine learning, there are a couple of steps that you need to follow.', 'start': 2256.7, 'duration': 5.226}, {'end': 2263.568, 'text': "All right, let's look at the steps.", 'start': 2262.266, 'duration': 1.302}, {'end': 2272.46, 'text': 'The first step is you define the objective of your problem and the second step is data gathering, which is followed by preparing your data,', 'start': 2264.272, 'duration': 8.188}, {'end': 2278.425, 'text': 'data exploration, building a model, model evaluation and finally, making your predictions.', 'start': 2272.46, 'duration': 5.965}, {'end': 2282.269, 'text': 'Now in order to understand the machine learning process.', 'start': 2279.506, 'duration': 2.763}, {'end': 2287.314, 'text': "let's assume that you've been given a problem that needs to be solved by using machine learning.", 'start': 2282.269, 'duration': 5.045}, {'end': 2295.429, 'text': 'So the problem that you need to solve is you need to predict the occurrence of rain in your local area by using machine learning.', 'start': 2288.5, 'duration': 6.929}, {'end': 2300.264, 'text': 'All right, so basically you need to predict the possibility of rain by studying the weather conditions.', 'start': 2295.982, 'duration': 4.282}, {'end': 2306.208, 'text': 'So what we did here is we basically looked at step number one which is define the objective of the problem.', 'start': 2300.705, 'duration': 5.503}, {'end': 2309.97, 'text': 'Now here you need to answer questions such as what are we trying to predict??', 'start': 2306.648, 'duration': 3.322}, {'end': 2315.533, 'text': 'Is our output going to be a continuous variable or is it going to be a discrete variable??', 'start': 2310.39, 'duration': 5.143}, {'end': 2321.176, 'text': 'These are the kind of questions that you need to answer in the first stage, which is defining the objective of the problem.', 'start': 2315.953, 'duration': 5.223}, {'end': 2324.459, 'text': 'Right, so yeah exactly what are the target features.', 'start': 2321.736, 'duration': 2.723}, {'end': 2332.245, 'text': 'So here you need to understand which is your target variable and what are the different predictor variables that you need in order to predict this outcome.', 'start': 2324.799, 'duration': 7.446}, {'end': 2338.471, 'text': "Right, so here our target variable will be basically a variable that can tell us whether it's going to rain or not.", 'start': 2332.826, 'duration': 5.645}, {'end': 2348.319, 'text': "Input data is we'll need data such as maybe the temperature on a particular day or the humidity level, the precipitation and so on.", 'start': 2339.469, 'duration': 8.85}, {'end': 2351.383, 'text': 'So you need to define the objective at this stage.', 'start': 2348.96, 'duration': 2.423}, {'end': 2355.568, 'text': 'So basically you have to form an idea of the problem at this stage.', 'start': 2351.743, 'duration': 3.825}, {'end': 2361.071, 'text': 'Another question that you need to ask yourself is what kind of problem are you solving?', 'start': 2356.789, 'duration': 4.282}, {'end': 2368.035, 'text': 'Is this a binary classification problem or is this a clustering problem or is this a regression problem?', 'start': 2361.491, 'duration': 6.544}, {'end': 2374.678, 'text': 'Now, a lot of you might not be familiar with the terms classification, clustering and regression in terms of machine learning.', 'start': 2368.395, 'duration': 6.283}, {'end': 2377.68, 'text': "Don't worry, I'll explain all of these terms in the upcoming slides.", 'start': 2374.698, 'duration': 2.982}, {'end': 2383.543, 'text': "All you need to understand at step one is you need to define how you're going to solve the problem.", 'start': 2378.2, 'duration': 5.343}, {'end': 2391.41, 'text': "You need to understand what sort of data you need to solve the problem, how you're going to approach the problem, what are you trying to predict,", 'start': 2383.903, 'duration': 7.507}, {'end': 2394.894, 'text': "what variables you'll need in order to predict the outcome, and so on.", 'start': 2391.41, 'duration': 3.484}, {'end': 2399.959, 'text': "Let's move on and look at our step number two, which is data gathering.", 'start': 2395.895, 'duration': 4.064}, {'end': 2407.352, 'text': 'Now, in this stage, you must be asking questions such as what kind of data is needed to solve this problem?', 'start': 2400.589, 'duration': 6.763}, {'end': 2409.573, 'text': 'And is this data available?', 'start': 2407.973, 'duration': 1.6}, {'end': 2414.335, 'text': 'And if it is available, from where can I get this data and how can I get the data?', 'start': 2409.833, 'duration': 4.502}, {'end': 2419.438, 'text': 'Data gathering is one of the most time consuming steps in machine learning process.', 'start': 2415.416, 'duration': 4.022}, {'end': 2423.681, 'text': "If you have to go manually and collect the data, it's going to take a lot of time.", 'start': 2420.218, 'duration': 3.463}, {'end': 2429.764, 'text': 'But lucky for us, there are a lot of resources online which provide data sets.', 'start': 2424.161, 'duration': 5.603}, {'end': 2433.707, 'text': 'All you have to do is web scraping or you just have to go ahead and download data.', 'start': 2429.784, 'duration': 3.923}, {'end': 2436.228, 'text': 'One of the websites I can tell you all about is Kaggle.', 'start': 2434.007, 'duration': 2.221}, {'end': 2440.591, 'text': "So if you're a beginner in machine learning, don't worry about data gathering and all of that.", 'start': 2436.248, 'duration': 4.343}, {'end': 2444.794, 'text': 'All you have to do is go to websites such as Kaggle and just download a data set.', 'start': 2440.951, 'duration': 3.843}, {'end': 2450.793, 'text': 'So, coming back to the problem that we are discussing, which is predicting the weather,', 'start': 2445.552, 'duration': 5.241}, {'end': 2459.214, 'text': 'the data needed for weather forecasting includes measures like humidity level, the temperature, the pressure, the locality,', 'start': 2450.793, 'duration': 8.421}, {'end': 2461.395, 'text': 'whether or not you live in a hill station.', 'start': 2459.214, 'duration': 2.181}], 'summary': 'Most ai technologies are based on machine learning, which involves training models with data to solve real-world problems.', 'duration': 531.764, 'max_score': 1929.631, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk1929631.jpg'}, {'end': 2429.764, 'src': 'embed', 'start': 2400.589, 'weight': 4, 'content': [{'end': 2407.352, 'text': 'Now, in this stage, you must be asking questions such as what kind of data is needed to solve this problem?', 'start': 2400.589, 'duration': 6.763}, {'end': 2409.573, 'text': 'And is this data available?', 'start': 2407.973, 'duration': 1.6}, {'end': 2414.335, 'text': 'And if it is available, from where can I get this data and how can I get the data?', 'start': 2409.833, 'duration': 4.502}, {'end': 2419.438, 'text': 'Data gathering is one of the most time consuming steps in machine learning process.', 'start': 2415.416, 'duration': 4.022}, {'end': 2423.681, 'text': "If you have to go manually and collect the data, it's going to take a lot of time.", 'start': 2420.218, 'duration': 3.463}, {'end': 2429.764, 'text': 'But lucky for us, there are a lot of resources online which provide data sets.', 'start': 2424.161, 'duration': 5.603}], 'summary': 'Data gathering is time-consuming; online resources provide datasets.', 'duration': 29.175, 'max_score': 2400.589, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk2400589.jpg'}, {'end': 2983.218, 'src': 'embed', 'start': 2953.249, 'weight': 3, 'content': [{'end': 2955.371, 'text': "So we'll go through this one by one.", 'start': 2953.249, 'duration': 2.122}, {'end': 2960.737, 'text': "We'll understand what supervised learning is first and then we look at the other two types.", 'start': 2955.972, 'duration': 4.765}, {'end': 2970.387, 'text': 'To define supervised learning, it is basically a technique in which we teach or train the machine by using the data which is well labeled.', 'start': 2961.277, 'duration': 9.11}, {'end': 2976.293, 'text': "Now in order to understand supervised learning, let's consider a small example.", 'start': 2970.847, 'duration': 5.446}, {'end': 2980.596, 'text': 'So as kids, we all needed guidance to solve math problems.', 'start': 2977.193, 'duration': 3.403}, {'end': 2983.218, 'text': 'A lot of us had trouble solving math problems.', 'start': 2980.916, 'duration': 2.302}], 'summary': 'Supervised learning teaches machines using labeled data.', 'duration': 29.969, 'max_score': 2953.249, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk2953249.jpg'}], 'start': 946.183, 'title': 'Ai applications, evolutionary stages, programming languages, machine learning importance & process', 'summary': 'Covers ai applications in netflix and gmail, highlighting machine learning and natural language processing, and discusses the evolutionary stages of ai. it also provides an overview of programming languages for ai, emphasizing python as the most effective. additionally, it emphasizes the importance of machine learning in handling vast data, citing the daily production of 2.5 quintillion bytes of data and the use of machine learning by top-tier companies. moreover, it explains the machine learning process and its types, highlighting the significance of labeled data in supervised learning.', 'chapters': [{'end': 1241.019, 'start': 946.183, 'title': 'Ai applications and evolutionary stages', 'summary': 'Discusses the applications of ai in netflix and gmail, highlighting the use of machine learning and natural language processing, and explains the three stages of ai - artificial narrow intelligence, artificial general intelligence, and artificial superintelligence.', 'duration': 294.836, 'highlights': ["Gmail uses AI and machine learning to classify emails as spam and non-spam based on words and correlations, such as 'lottery' and 'full refund', and separate them into different sections, demonstrating practical AI applications.", 'The chapter explains the three stages of AI - artificial narrow intelligence, artificial general intelligence, and artificial superintelligence - and provides examples such as Alexa, Google search engine, and AlphaGo to illustrate each stage.', 'It emphasizes that the current stage of AI is artificial narrow intelligence, displaying the limitations of machines in reasoning and human-like abilities despite their strong processing capabilities.']}, {'end': 1628.472, 'start': 1241.019, 'title': 'Best programming languages for ai', 'summary': 'Provides an overview of various programming languages for ai, highlighting python as the most effective with its simplicity, extensive libraries, and widespread usage, followed by r and java, and mentions lisp and prolog as older but still relevant options.', 'duration': 387.453, 'highlights': ['Python is the most effective language for AI with its simplicity, extensive libraries, and widespread usage, making it the best choice for artificial intelligence over R, Java, Lisp, Prolog, and other languages.', 'R is also an effective language for AI with its similarity to English, ease of learning, and numerous libraries supporting statistics, data science, and machine learning.', 'Java is a good choice for AI development with its benefits such as easy debugging, simplified work with large-scale projects, good user interaction, graphical representation of data, and the standard widget toolkit for making graphs and interfaces.', 'Lisp is the oldest and most suited language for AI development, invented by John McCarthy, the father of artificial intelligence, and is known for its capability of processing symbolic information, excellent prototyping capabilities, and dynamic object creation.', 'Prolog is frequently used in knowledge base and expert systems, widely utilized in medical projects, and offers powerful features such as pattern matching, tree-based data structuring, and automatic backtracking.']}, {'end': 2377.68, 'start': 1628.472, 'title': 'Importance of machine learning in ai', 'summary': 'Discusses the significance of machine learning in handling the vast amounts of data generated daily, stating that 2.5 quintillion bytes of data are produced daily and 1.7 mb of data will be created every second for every person on earth by 2020, and highlights that top-tier companies like netflix and amazon utilize machine learning models to identify profitable opportunities and mitigate risks.', 'duration': 749.208, 'highlights': ['2.5 quintillion bytes of data are produced daily and 1.7 MB of data will be created every second for every person on Earth by 2020', 'Top-tier companies like Netflix and Amazon utilize machine learning models to identify profitable opportunities and mitigate risks', 'Machine learning is used to solve problems and find solutions to complex tasks faced by organizations']}, {'end': 2768.202, 'start': 2378.2, 'title': 'Machine learning process overview', 'summary': 'Discusses the machine learning process, emphasizing the key steps including defining the problem, data gathering from online resources such as kaggle, data cleaning, exploratory data analysis, and building a machine learning model by splitting the dataset into training and testing data.', 'duration': 390.002, 'highlights': ['The most time-consuming step in machine learning is data gathering, with online resources like Kaggle providing datasets and web scraping options, saving significant time. (Relevancy Score: 5)', 'Data cleaning is identified as the most difficult and time-consuming step in machine learning, as per a survey where 80% of data scientists found it challenging. (Relevancy Score: 4)', 'Exploratory Data Analysis involves understanding patterns and correlations between variables, crucial for deriving insights and mapping the solution to the problem. (Relevancy Score: 3)', 'Building a machine learning model involves splitting the dataset into training and testing data, where the training data, comprising 80% of the dataset, is used to train the model for better prediction outcomes. (Relevancy Score: 2)']}, {'end': 3188.475, 'start': 2768.822, 'title': 'Machine learning process & types', 'summary': 'Explains the machine learning process, including model building, evaluation, and optimization, as well as the types of machine learning: supervised, unsupervised, and reinforcement learning, emphasizing the significance of labeled data in supervised learning.', 'duration': 419.653, 'highlights': ['Supervised learning involves training the machine using well-labeled data, guiding the machine to understand patterns, and classifying input data into labeled output, essential for the training data set.', 'Unsupervised learning involves training by using unlabeled data, allowing the model to identify patterns and differences in data without guidance, thus functioning without labeled input or output data.', 'Model evaluation and optimization stage involves testing the efficiency of the model using the testing data set, calculating accuracy, and implementing improvements such as parameter tuning and cross-validation methods to enhance performance.']}], 'duration': 2242.292, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk946183.jpg', 'highlights': ['Python is the most effective language for AI with its simplicity, extensive libraries, and widespread usage, making it the best choice for artificial intelligence over R, Java, Lisp, Prolog, and other languages.', "Gmail uses AI and machine learning to classify emails as spam and non-spam based on words and correlations, such as 'lottery' and 'full refund', and separate them into different sections, demonstrating practical AI applications.", 'Top-tier companies like Netflix and Amazon utilize machine learning models to identify profitable opportunities and mitigate risks.', 'Supervised learning involves training the machine using well-labeled data, guiding the machine to understand patterns, and classifying input data into labeled output, essential for the training data set.', 'The most time-consuming step in machine learning is data gathering, with online resources like Kaggle providing datasets and web scraping options, saving significant time.']}, {'end': 4576.847, 'segs': [{'end': 3255.609, 'src': 'embed', 'start': 3235.587, 'weight': 1, 'content': [{'end': 3248.553, 'text': 'It is a part of machine learning where an agent is put in an environment and he learns to behave in this environment by performing certain actions and observing the reward which it gets from those actions.', 'start': 3235.587, 'duration': 12.966}, {'end': 3255.609, 'text': 'To understand what reinforcement learning is, imagine that you were dropped off at an isolated island.', 'start': 3249.4, 'duration': 6.209}], 'summary': 'Reinforcement learning teaches agents to behave in an environment through actions and rewards.', 'duration': 20.022, 'max_score': 3235.587, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3235587.jpg'}, {'end': 3439.421, 'src': 'embed', 'start': 3409.949, 'weight': 2, 'content': [{'end': 3412.071, 'text': "In unsupervised learning, there's no supervision.", 'start': 3409.949, 'duration': 2.122}, {'end': 3415.254, 'text': "Again, in reinforcement learning, there's no supervision at all.", 'start': 3412.491, 'duration': 2.763}, {'end': 3423.336, 'text': 'Now what is the approach to solve problems by using supervised, unsupervised, and reinforcement learning? In supervised learning, it is simple.', 'start': 3415.894, 'duration': 7.442}, {'end': 3426.717, 'text': 'You have to map the labeled input to the known output.', 'start': 3423.616, 'duration': 3.101}, {'end': 3429.078, 'text': 'The machine knows what the output looks like.', 'start': 3427.257, 'duration': 1.821}, {'end': 3431.639, 'text': "So you're just labeling the input to the output.", 'start': 3429.378, 'duration': 2.261}, {'end': 3436.5, 'text': "In unsupervised learning, you're going to understand the patterns and discover the output.", 'start': 3432.219, 'duration': 4.281}, {'end': 3439.421, 'text': 'Here you have no clue about what the input is.', 'start': 3436.96, 'duration': 2.461}], 'summary': 'Supervised learning maps labeled input to known output, unsupervised learning understands patterns to discover the output.', 'duration': 29.472, 'max_score': 3409.949, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3409949.jpg'}, {'end': 3497.192, 'src': 'embed', 'start': 3473.834, 'weight': 3, 'content': [{'end': 3482.341, 'text': 'Popular algorithms under supervised learning include linear regression, logistic regression, support vector machines, k-nearest, neighbor,', 'start': 3473.834, 'duration': 8.507}, {'end': 3483.642, 'text': 'naive bias and so on.', 'start': 3482.341, 'duration': 1.301}, {'end': 3489.706, 'text': 'Under unsupervised learning, we have the famous k-means clustering method, c-means, and all of that.', 'start': 3484.102, 'duration': 5.604}, {'end': 3493.93, 'text': 'Under reinforcement learning, we have the famous Q-learning algorithm.', 'start': 3490.207, 'duration': 3.723}, {'end': 3497.192, 'text': "I'll be discussing these algorithms in the upcoming slides.", 'start': 3494.35, 'duration': 2.842}], 'summary': 'Supervised learning: linear regression, logistic regression, support vector machines, k-nearest neighbor, naive bias. unsupervised learning: k-means clustering, c-means. reinforcement learning: q-learning. algorithms to be discussed in upcoming slides.', 'duration': 23.358, 'max_score': 3473.834, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3473834.jpg'}, {'end': 4043.087, 'src': 'heatmap', 'start': 3513.18, 'weight': 0.854, 'content': [{'end': 3518.822, 'text': 'In machine learning, all the problems can be classified into three types.', 'start': 3513.18, 'duration': 5.642}, {'end': 3523.884, 'text': 'Every problem that is approached in machine learning can be put into one of these three categories.', 'start': 3518.942, 'duration': 4.942}, {'end': 3527.294, 'text': 'Okay, so the first type is known as regression.', 'start': 3524.671, 'duration': 2.623}, {'end': 3530.237, 'text': 'Then we have classification and clustering.', 'start': 3527.874, 'duration': 2.363}, {'end': 3533.04, 'text': "So first let's look at regression type of problems.", 'start': 3530.737, 'duration': 2.303}, {'end': 3537.785, 'text': 'So in this type of problem, the output is always a continuous quantity.', 'start': 3533.58, 'duration': 4.205}, {'end': 3544.672, 'text': 'For example, if you want to predict the speed of a car given the distance, it is a regression problem.', 'start': 3538.345, 'duration': 6.327}, {'end': 3549.455, 'text': 'Now, a lot of you might not be very aware of what exactly a continuous quantity is.', 'start': 3545.152, 'duration': 4.303}, {'end': 3554.219, 'text': 'A continuous quantity is any quantity that can have an infinite range of values.', 'start': 3549.936, 'duration': 4.283}, {'end': 3556.841, 'text': 'For example, the weight of a person.', 'start': 3555.299, 'duration': 1.542}, {'end': 3560.964, 'text': 'It is a continuous quantity because our weight can be 50, 50.1, 50.001, 50.0021, 50.0321 and so on.', 'start': 3557.261, 'duration': 3.703}, {'end': 3571.972, 'text': 'it can have an infinite range of values, correct?', 'start': 3568.909, 'duration': 3.063}, {'end': 3579.698, 'text': 'So in the type of problems that you have to predict a continuous quantity, you make use of regression algorithms, all right?', 'start': 3572.112, 'duration': 7.586}, {'end': 3585.082, 'text': 'So regression problems can be solved by using supervised learning algorithms like linear regression.', 'start': 3579.958, 'duration': 5.124}, {'end': 3587.571, 'text': 'Next we have classification.', 'start': 3586.11, 'duration': 1.461}, {'end': 3592.753, 'text': 'Now in this type of problems, the output is always a categorical value.', 'start': 3588.071, 'duration': 4.682}, {'end': 3598.796, 'text': 'Now, when I say categorical value, it can be values such as the gender of a person.', 'start': 3593.233, 'duration': 5.563}, {'end': 3600.957, 'text': 'is a categorical value okay?', 'start': 3598.796, 'duration': 2.161}, {'end': 3605.358, 'text': 'Now, classifying emails into two classes, like spam and non-spam,', 'start': 3601.477, 'duration': 3.881}, {'end': 3613.481, 'text': 'is a classification problem that can be solved by using supervised learning classification algorithms like support vector machines, naive bias,', 'start': 3605.358, 'duration': 8.123}, {'end': 3616.362, 'text': 'logistic regression, k-nearest neighbor and so on.', 'start': 3613.481, 'duration': 2.881}, {'end': 3622.624, 'text': 'So again, the main aim in classification is to compute the category of the data.', 'start': 3617.782, 'duration': 4.842}, {'end': 3632.764, 'text': 'Coming to clustering problems, this type of problem involves assigning the input into two or more clusters based on feature similarity.', 'start': 3624.06, 'duration': 8.704}, {'end': 3637.746, 'text': 'As soon as I read this sentence, you should understand that this is unsupervised learning,', 'start': 3633.404, 'duration': 4.342}, {'end': 3643.508, 'text': "because you don't have enough data about your input and the only option that you have is to form clusters.", 'start': 3637.746, 'duration': 5.762}, {'end': 3650.391, 'text': "Categories are formed only when you know that your data is of two type, your input data is labeled and it's of two types.", 'start': 3644.568, 'duration': 5.823}, {'end': 3652.512, 'text': "So it's going to be a classification problem.", 'start': 3650.751, 'duration': 1.761}, {'end': 3657.694, 'text': "But when a clustering problem happens, when you don't have much information about your input,", 'start': 3652.952, 'duration': 4.742}, {'end': 3669.12, 'text': 'all you have to do is you have to find patterns and you have to understand that data points which are similar are clustered into one group and data points which are different from the first group are clustered into another group.', 'start': 3657.694, 'duration': 11.426}, {'end': 3671.161, 'text': "That's what clustering is.", 'start': 3670.06, 'duration': 1.101}, {'end': 3674.001, 'text': 'An example is in Netflix.', 'start': 3671.901, 'duration': 2.1}, {'end': 3683.184, 'text': 'what happens is Netflix clusters their users into similar groups based on their interest, based on their age, geography and so on.', 'start': 3674.001, 'duration': 9.183}, {'end': 3688.006, 'text': 'This can be done by using unsupervised learning algorithms like k-means.', 'start': 3684.225, 'duration': 3.781}, {'end': 3694.408, 'text': 'So guys, those were the three categories of problems that can be solved by using machine learning.', 'start': 3689.606, 'duration': 4.802}, {'end': 3699.796, 'text': "So basically what I'm trying to say is all the problems will fall into one of these categories.", 'start': 3694.952, 'duration': 4.844}, {'end': 3705.501, 'text': "So any problem that you give to a machine learning model, it'll fall into one of these categories, okay?", 'start': 3700.236, 'duration': 5.265}, {'end': 3707.923, 'text': 'Now to make things a little more interesting,', 'start': 3706.001, 'duration': 1.922}, {'end': 3720.894, 'text': "I have collected real world data sets from online resources and what we're gonna do is we're going to try and understand if this is a regression problem or a clustering problem or a classification problem.", 'start': 3707.923, 'duration': 12.971}, {'end': 3730.668, 'text': 'okay?. Now the problem statement in here is to study the house sales data set and build a machine learning model that predicts the house pricing index.', 'start': 3720.894, 'duration': 9.774}, {'end': 3738.357, 'text': 'Now, the most important thing you need to understand when you read a problem statement is you need to understand what is your target variable.', 'start': 3731.289, 'duration': 7.068}, {'end': 3741.08, 'text': "what are the possible predictor variables that you'll need?", 'start': 3738.357, 'duration': 2.723}, {'end': 3744.796, 'text': 'The first thing you should look at is your target variable.', 'start': 3741.733, 'duration': 3.063}, {'end': 3749.319, 'text': 'If you want to understand if this is a classification regression or clustering problem,', 'start': 3744.956, 'duration': 4.363}, {'end': 3753.423, 'text': "look at your target variable or your output variable that you're supposed to predict.", 'start': 3749.319, 'duration': 4.104}, {'end': 3756.405, 'text': "Here you're supposed to predict the house pricing index.", 'start': 3753.963, 'duration': 2.442}, {'end': 3760.428, 'text': 'Now house pricing index is obviously a continuous quantity.', 'start': 3756.905, 'duration': 3.523}, {'end': 3764.592, 'text': 'So as soon as you understand that, you know that this is a regression problem.', 'start': 3760.949, 'duration': 3.643}, {'end': 3770.736, 'text': 'So for this you can make use of the linear regression algorithm and you can predict the house pricing index.', 'start': 3765.672, 'duration': 5.064}, {'end': 3774.366, 'text': 'Linear regression is a regression algorithm.', 'start': 3771.684, 'duration': 2.682}, {'end': 3776.828, 'text': 'It is a supervised learning algorithm.', 'start': 3774.786, 'duration': 2.042}, {'end': 3779.41, 'text': "We'll discuss more about it in the further slides.", 'start': 3777.428, 'duration': 1.982}, {'end': 3781.371, 'text': "Let's look at our next problem statement.", 'start': 3779.73, 'duration': 1.641}, {'end': 3790.978, 'text': 'Here you have to study a bank credit data set and make a decision about whether to approve the loan of an applicant based on his profile.', 'start': 3781.952, 'duration': 9.026}, {'end': 3800.105, 'text': 'Now what is your output variable over here? Your output variable is to predict whether you can approve the loan of an applicant or not.', 'start': 3791.379, 'duration': 8.726}, {'end': 3803.772, 'text': 'So obviously your output is going to be categorical.', 'start': 3800.729, 'duration': 3.043}, {'end': 3805.914, 'text': "It's either going to be yes or no.", 'start': 3804.252, 'duration': 1.662}, {'end': 3809.377, 'text': 'Yes is basically approve loan, no is reject loan.', 'start': 3806.374, 'duration': 3.003}, {'end': 3814.342, 'text': 'So here itself you understand that this is a classification problem.', 'start': 3810.398, 'duration': 3.944}, {'end': 3822.57, 'text': 'So you can make use of algorithms like KNN algorithm or you can make use of support vector machines in order to do this.', 'start': 3815.423, 'duration': 7.147}, {'end': 3831.022, 'text': 'So support vector machines and the KNN, which is k-nearest neighbor algorithms, are basically supervised learning algorithms.', 'start': 3823.354, 'duration': 7.668}, {'end': 3833.184, 'text': "We'll talk more about that in the upcoming slides.", 'start': 3831.082, 'duration': 2.102}, {'end': 3835.727, 'text': 'Moving on to our next problem statement.', 'start': 3833.865, 'duration': 1.862}, {'end': 3843.855, 'text': 'Here, the problem statement is to cluster a set of movies as either good or average based on their social media outreach.', 'start': 3836.247, 'duration': 7.608}, {'end': 3848.046, 'text': 'Now if you look properly, your clue is in the question itself.', 'start': 3844.705, 'duration': 3.341}, {'end': 3854.269, 'text': 'The first line itself says to cluster a set of movies as either good or average.', 'start': 3848.527, 'duration': 5.742}, {'end': 3862.453, 'text': 'Now, guys, whenever you have a problem statement, that is asking you to group the data set into different groups or to form different,', 'start': 3854.729, 'duration': 7.724}, {'end': 3865.514, 'text': "different clusters, it's obviously a clustering problem.", 'start': 3862.453, 'duration': 3.061}, {'end': 3870.597, 'text': 'Here you can make use of the k-means clustering algorithm and you can form two clusters.', 'start': 3866.214, 'duration': 4.383}, {'end': 3876.16, 'text': 'One will contain the popular movies and the other will contain the non-popular movies.', 'start': 3871.117, 'duration': 5.043}, {'end': 3883.384, 'text': 'These are small examples of how you can use machine learning to solve clustering problems, regression and classification problems.', 'start': 3876.62, 'duration': 6.764}, {'end': 3887.487, 'text': 'The key is you need to identify the type of problem first.', 'start': 3883.945, 'duration': 3.542}, {'end': 3892.65, 'text': "Now let's move on and discuss the different types of machine learning algorithms.", 'start': 3888.107, 'duration': 4.543}, {'end': 3897.758, 'text': "Right, so we're going to start by discussing the different supervised learning algorithms.", 'start': 3893.797, 'duration': 3.961}, {'end': 3907.842, 'text': "So, to give you a quick overview, we'll be discussing the linear regression logistic regression decision tree random forest naive Bayes classifier,", 'start': 3898.459, 'duration': 9.383}, {'end': 3910.463, 'text': 'support vector machines and k-nearest neighbor.', 'start': 3907.842, 'duration': 2.621}, {'end': 3913.388, 'text': "We'll be discussing these seven algorithms.", 'start': 3911.387, 'duration': 2.001}, {'end': 3917.848, 'text': "So without any further delay, let's look at linear regression first.", 'start': 3914.168, 'duration': 3.68}, {'end': 3921.909, 'text': 'Now, what exactly is a linear regression algorithm?', 'start': 3919.069, 'duration': 2.84}, {'end': 3922.81, 'text': 'So guys,', 'start': 3922.489, 'duration': 0.321}, {'end': 3933.732, 'text': 'linear regression is basically a supervised learning algorithm that is used to predict a continuous dependent variable y based on the values of independent variable x.', 'start': 3922.81, 'duration': 10.922}, {'end': 3941.677, 'text': "The important thing to note here is that the dependent variable y, the variable that you're trying to predict,", 'start': 3935.312, 'duration': 6.365}, {'end': 3944.339, 'text': 'is always going to be a continuous variable.', 'start': 3941.677, 'duration': 2.662}, {'end': 3950.685, 'text': 'But the independent variable x, which is basically the predictor variables.', 'start': 3944.9, 'duration': 5.785}, {'end': 3956.37, 'text': "these are the variables that you'll be using to predict your output variable, which is nothing but your dependent variable.", 'start': 3950.685, 'duration': 5.685}, {'end': 3962.936, 'text': 'So your independent variables or your predictor variables can either be continuous or discrete.', 'start': 3956.972, 'duration': 5.964}, {'end': 3965.819, 'text': 'There is no such restriction over here.', 'start': 3962.976, 'duration': 2.843}, {'end': 3969.021, 'text': 'They can be either continuous variables or they can be discrete variables.', 'start': 3965.839, 'duration': 3.182}, {'end': 3973.801, 'text': "Now again, I'll tell you what a continuous variable is in case you've forgotten.", 'start': 3969.918, 'duration': 3.883}, {'end': 3977.584, 'text': 'It is a variable that has infinite number of possibilities.', 'start': 3974.341, 'duration': 3.243}, {'end': 3981.107, 'text': "So I give you an example of a person's weight.", 'start': 3978.064, 'duration': 3.043}, {'end': 3989.213, 'text': 'It can be 160 pounds or they can weigh 160.11 pounds or 160.1134 pounds and so on.', 'start': 3981.187, 'duration': 8.026}, {'end': 3994.057, 'text': 'So the number of possibilities for weight is limitless.', 'start': 3989.373, 'duration': 4.684}, {'end': 3996.519, 'text': 'And this is exactly what a continuous variable is.', 'start': 3994.417, 'duration': 2.102}, {'end': 4004.943, 'text': "Now in order to understand linear regression, let's assume that you want to predict the price of a stock over a period of time.", 'start': 3997.239, 'duration': 7.704}, {'end': 4014.328, 'text': 'For such a problem, you can make use of linear regression by studying the relationship between the dependent variable, which is the stock price,', 'start': 4005.983, 'duration': 8.345}, {'end': 4016.569, 'text': 'and the independent variable, which is the time.', 'start': 4014.328, 'duration': 2.241}, {'end': 4021.431, 'text': "You're trying to predict the stock price over a period of time.", 'start': 4017.229, 'duration': 4.202}, {'end': 4026.814, 'text': "So basically you're going to check how the price of a stock varies over a period of time.", 'start': 4022.012, 'duration': 4.802}, {'end': 4032.861, 'text': 'So your stock price is going to be your dependent variable or your output variable,', 'start': 4027.438, 'duration': 5.423}, {'end': 4037.424, 'text': 'and the time is going to be your predictor variable or your independent variable.', 'start': 4032.861, 'duration': 4.563}, {'end': 4039.805, 'text': "Let's not confuse it anymore.", 'start': 4037.904, 'duration': 1.901}, {'end': 4043.087, 'text': 'Your dependent variable is your output variable.', 'start': 4040.146, 'duration': 2.941}], 'summary': 'Machine learning problems can be categorized into regression, classification, and clustering, each with specific algorithms for solving them.', 'duration': 529.907, 'max_score': 3513.18, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3513180.jpg'}, {'end': 3650.391, 'src': 'embed', 'start': 3624.06, 'weight': 0, 'content': [{'end': 3632.764, 'text': 'Coming to clustering problems, this type of problem involves assigning the input into two or more clusters based on feature similarity.', 'start': 3624.06, 'duration': 8.704}, {'end': 3637.746, 'text': 'As soon as I read this sentence, you should understand that this is unsupervised learning,', 'start': 3633.404, 'duration': 4.342}, {'end': 3643.508, 'text': "because you don't have enough data about your input and the only option that you have is to form clusters.", 'start': 3637.746, 'duration': 5.762}, {'end': 3650.391, 'text': "Categories are formed only when you know that your data is of two type, your input data is labeled and it's of two types.", 'start': 3644.568, 'duration': 5.823}], 'summary': 'Clustering involves unsupervised learning to form clusters based on feature similarity.', 'duration': 26.331, 'max_score': 3624.06, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3624060.jpg'}, {'end': 4420.852, 'src': 'embed', 'start': 4392.63, 'weight': 4, 'content': [{'end': 4398.753, 'text': 'So basically this shows that we have around 12, 000 rows and 31 columns in our data set.', 'start': 4392.63, 'duration': 6.123}, {'end': 4402.655, 'text': 'Now the 31 columns basically represent the predictor variables.', 'start': 4399.193, 'duration': 3.462}, {'end': 4408.017, 'text': 'So you can see that we have 31 predictor variables in order to predict the weather conditions on a particular day.', 'start': 4403.035, 'duration': 4.982}, {'end': 4412.302, 'text': 'So guys, the main aim in this problem statement is weather forecasting.', 'start': 4408.477, 'duration': 3.825}, {'end': 4415.986, 'text': "We're going to predict the weather by using a set of predictor variables.", 'start': 4412.322, 'duration': 3.664}, {'end': 4420.852, 'text': 'So these are the different types of predictor variables that we have.', 'start': 4416.948, 'duration': 3.904}], 'summary': 'Data set has 12,000 rows and 31 predictor variables for weather forecasting.', 'duration': 28.222, 'max_score': 4392.63, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk4392630.jpg'}], 'start': 3188.815, 'title': 'Types and problems of machine learning', 'summary': 'Discusses three types of machine learning: supervised learning, unsupervised learning, and reinforcement learning, and popular algorithms under supervised, unsupervised, and reinforcement learning, followed by an explanation of regression, classification, and clustering problems in machine learning, with examples and real-world datasets. additionally, it includes a linear regression demo in python using a dataset containing around 12,000 rows and 31 predictor variables.', 'chapters': [{'end': 3314.004, 'start': 3188.815, 'title': 'Types of machine learning', 'summary': 'Discusses three types of machine learning: supervised learning, unsupervised learning, and reinforcement learning. unsupervised learning forms clusters based on feature similarity, while reinforcement learning involves an agent learning to behave in an environment by observing rewards from actions.', 'duration': 125.189, 'highlights': ['Unsupervised learning forms clusters based on feature similarity, and the machine is fed with unlabeled data to discover patterns and output on its own.', 'Reinforcement learning involves an agent learning to behave in an environment by performing actions and observing the rewards obtained, and it is used in advanced machine learning areas such as self-driving cars and AlphaGo.', 'Reinforcement learning is compared to being dropped off at an isolated island where the agent must learn by observing and performing actions that result in rewards.']}, {'end': 3472.089, 'start': 3314.384, 'title': 'Types of machine learning', 'summary': 'Explains the types of machine learning including supervised, unsupervised, and reinforcement learning, the problems they solve, the type of data used, the training involved, and the approach to problem-solving, with reinforcement learning being totally based on the concept of trial and error.', 'duration': 157.705, 'highlights': ['Reinforcement learning is totally based on the concept of trial and error, where the agent interacts with the environment, discovers errors or rewards based on his actions.', 'Supervised learning involves mapping the labeled input to the known output and relies on labeled data for training.', 'Unsupervised learning focuses on understanding patterns and discovering the output without labeled input data.']}, {'end': 3814.342, 'start': 3473.834, 'title': 'Types of machine learning problems', 'summary': 'Covers popular algorithms under supervised, unsupervised, and reinforcement learning, followed by an explanation of regression, classification, and clustering problems in machine learning, with examples and real-world datasets.', 'duration': 340.508, 'highlights': ['The chapter covers popular algorithms under supervised, unsupervised, and reinforcement learning', 'Explanation of regression, classification, and clustering problems in machine learning', 'Examples of real-world datasets and problem statements']}, {'end': 4211.663, 'start': 3815.423, 'title': 'Machine learning algorithms overview', 'summary': 'Discusses the application of machine learning algorithms such as knn and support vector machines in clustering, regression, and classification problems, including an overview of linear regression and its mathematical representation.', 'duration': 396.24, 'highlights': ['The chapter discusses the application of machine learning algorithms such as KNN and support vector machines in clustering, regression, and classification problems.', 'The chapter provides an overview of linear regression and its mathematical representation.', 'The chapter explains the concept of supervised learning algorithms, including linear regression, logistic regression, decision tree, random forest, naive Bayes classifier, support vector machines, and k-nearest neighbor.']}, {'end': 4576.847, 'start': 4213.05, 'title': 'Understanding linear regression demo in python', 'summary': 'Demonstrates a linear regression demo in python to predict the maximum temperature based on the minimum temperature, using a dataset containing around 12,000 rows and 31 predictor variables, showing a clear linear relationship between the two temperatures and exploring the average maximum temperature range.', 'duration': 363.797, 'highlights': ['The dataset contains around 12,000 rows and 31 columns, representing predictor variables, for weather forecasting.', 'A clear linear relationship is observed between the minimum temperature (independent variable) and maximum temperature (dependent variable), making it suitable for linear regression.', 'Exploratory data analysis includes examining the average maximum temperature, which lies between 25 and 35.']}], 'duration': 1388.032, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk3188815.jpg', 'highlights': ['Unsupervised learning forms clusters based on feature similarity', 'Reinforcement learning involves an agent learning to behave in an environment', 'Supervised learning involves mapping the labeled input to the known output', 'The chapter covers popular algorithms under supervised, unsupervised, and reinforcement learning', 'The dataset contains around 12,000 rows and 31 columns, representing predictor variables']}, {'end': 6853.981, 'segs': [{'end': 4676.62, 'src': 'embed', 'start': 4634.589, 'weight': 1, 'content': [{'end': 4640.693, 'text': 'So basically we had 30 predictor variables and we had one target variable, which is your maximum temperature.', 'start': 4634.589, 'duration': 6.104}, {'end': 4648.518, 'text': "So what I'm doing here is I'm only considering these two variables because I want to show you exactly how linear regression works.", 'start': 4641.904, 'duration': 6.614}, {'end': 4657.267, 'text': "So here what I'm doing is I'm basically extracting only these two variables from our data set, storing it in X and Y.", 'start': 4649.261, 'duration': 8.006}, {'end': 4659.148, 'text': "After that, I'm performing data splicing.", 'start': 4657.267, 'duration': 1.881}, {'end': 4663.111, 'text': "So here I'm basically splitting the data into training and testing data.", 'start': 4659.688, 'duration': 3.423}, {'end': 4671.296, 'text': 'And remember one point that I am assigning 20% of the data to our testing data set and the remaining 80% is assigned for training.', 'start': 4663.571, 'duration': 7.725}, {'end': 4673.758, 'text': "That's how training works.", 'start': 4672.117, 'duration': 1.641}, {'end': 4676.62, 'text': 'We assign maximum data set for training.', 'start': 4674.278, 'duration': 2.342}], 'summary': 'Performed linear regression with 30 predictor variables and 1 target variable, splitting data into 80% training and 20% testing.', 'duration': 42.031, 'max_score': 4634.589, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk4634589.jpg'}, {'end': 5748.607, 'src': 'embed', 'start': 5722.67, 'weight': 3, 'content': [{'end': 5728.274, 'text': "Then I'll show you a practical demo where we'll use multiple classification algorithms to solve the same problem.", 'start': 5722.67, 'duration': 5.604}, {'end': 5733.738, 'text': "Okay, and we'll also calculate the accuracy and see which classification algorithm is doing the best.", 'start': 5728.815, 'duration': 4.923}, {'end': 5737.3, 'text': "Now the next algorithm I'm gonna talk about is decision tree.", 'start': 5734.278, 'duration': 3.022}, {'end': 5742.684, 'text': "Decision tree is one of my favorite algorithms because it's very simple to understand how a decision tree works.", 'start': 5737.62, 'duration': 5.064}, {'end': 5748.607, 'text': 'So guys, before this, we discussed linear regression, which is a regression algorithm.', 'start': 5743.823, 'duration': 4.784}], 'summary': 'Demonstrating multiple classification algorithms, comparing accuracy, favoring decision tree.', 'duration': 25.937, 'max_score': 5722.67, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk5722670.jpg'}, {'end': 6052.926, 'src': 'embed', 'start': 6023.963, 'weight': 4, 'content': [{'end': 6029.208, 'text': 'Now the ID3 algorithm has around six defined steps in order to build a decision tree.', 'start': 6023.963, 'duration': 5.245}, {'end': 6033.342, 'text': 'So the first step is you will select the best attribute.', 'start': 6029.961, 'duration': 3.381}, {'end': 6039.763, 'text': 'Now what do you mean by the best attribute? So attribute is nothing but the predictor variable over here.', 'start': 6033.722, 'duration': 6.041}, {'end': 6041.784, 'text': "So you'll select the best predictor variable.", 'start': 6039.823, 'duration': 1.961}, {'end': 6043.544, 'text': "Let's call it A.", 'start': 6042.144, 'duration': 1.4}, {'end': 6048.065, 'text': "After that you'll assign this A as a decision variable for the root node.", 'start': 6043.544, 'duration': 4.521}, {'end': 6052.926, 'text': "Basically you'll assign this predictor variable A at the root node.", 'start': 6048.505, 'duration': 4.421}], 'summary': 'The id3 algorithm has six steps to build a decision tree, starting with selecting the best predictor variable.', 'duration': 28.963, 'max_score': 6023.963, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk6023963.jpg'}, {'end': 6333.297, 'src': 'heatmap', 'start': 6147.665, 'weight': 0.773, 'content': [{'end': 6154.787, 'text': 'Now, the best attribute is the one that separates the data into different classes most effectively,', 'start': 6147.665, 'duration': 7.122}, {'end': 6159.488, 'text': 'or it is basically a feature that best splits the data set right?', 'start': 6154.787, 'duration': 4.701}, {'end': 6168.531, 'text': 'Now the next question in your head must be how do I decide which variable or which feature best splits the data?', 'start': 6159.928, 'duration': 8.603}, {'end': 6172.352, 'text': 'To do this, there are two important measures right?', 'start': 6169.271, 'duration': 3.081}, {'end': 6176.353, 'text': "There's something known as information gain and there's something known as entropy.", 'start': 6172.572, 'duration': 3.781}, {'end': 6182.071, 'text': 'Now guys, in order to understand information gain and entropy, we look at a simple problem statement.', 'start': 6177.128, 'duration': 4.943}, {'end': 6187.315, 'text': 'This data set represents the speed of a car based on certain parameters.', 'start': 6182.752, 'duration': 4.563}, {'end': 6196.181, 'text': 'So our problem statement here is to study the data set and create a decision tree that classifies the speed of a car as either slow or fast.', 'start': 6187.875, 'duration': 8.306}, {'end': 6205.107, 'text': 'So our predictor variables here are road type, obstruction, and speed limit, and our response variable or our output variable is speed.', 'start': 6196.961, 'duration': 8.146}, {'end': 6210.531, 'text': "So we'd be building a decision tree using these variables in order to predict the speed of a car.", 'start': 6205.825, 'duration': 4.706}, {'end': 6212.273, 'text': 'Now, like I mentioned earlier,', 'start': 6211.052, 'duration': 1.221}, {'end': 6220.464, 'text': 'we must first begin by deciding a variable that best splits the data set and assign that particular variable to the root node,', 'start': 6212.273, 'duration': 8.191}, {'end': 6223.407, 'text': 'and repeat the same thing for other nodes as well.', 'start': 6220.464, 'duration': 2.943}, {'end': 6223.648, 'text': 'all right?', 'start': 6223.407, 'duration': 0.241}, {'end': 6228.967, 'text': 'So step one, like we discussed earlier, is to select the best attribute A.', 'start': 6224.426, 'duration': 4.541}, {'end': 6232.068, 'text': 'Now, how do you know which variable best separates the data?', 'start': 6228.967, 'duration': 3.101}, {'end': 6238.429, 'text': 'The variable with the highest information gain best divides the data into the desired output classes.', 'start': 6232.568, 'duration': 5.861}, {'end': 6240.989, 'text': "First of all, we'll calculate two measures.", 'start': 6239.049, 'duration': 1.94}, {'end': 6243.99, 'text': "We'll calculate the entropy and the information gain.", 'start': 6241.389, 'duration': 2.601}, {'end': 6249.231, 'text': 'Now this is where I tell you what exactly entropy is and what exactly information gain is.', 'start': 6244.55, 'duration': 4.681}, {'end': 6257.573, 'text': 'Now entropy is basically used to measure the impurity or the uncertainty present in the data.', 'start': 6250.382, 'duration': 7.191}, {'end': 6261.62, 'text': 'It is used to decide how a decision tree can split the data.', 'start': 6258.154, 'duration': 3.466}, {'end': 6267.577, 'text': 'Information gain on the other hand is the most significant measure which is used to build a decision tree.', 'start': 6262.236, 'duration': 5.341}, {'end': 6274.139, 'text': 'It indicates how much information a particular variable gives us about the final outcome.', 'start': 6268.197, 'duration': 5.942}, {'end': 6280.88, 'text': 'So information gain is important because it is used to choose a variable that best splits the data at each node for a decision tree.', 'start': 6274.559, 'duration': 6.321}, {'end': 6285.801, 'text': 'Now the variable with the highest information gain will be used to split the data at the root node.', 'start': 6281.32, 'duration': 4.481}, {'end': 6289.302, 'text': 'Now in our data set, there are four observations.', 'start': 6286.301, 'duration': 3.001}, {'end': 6296.843, 'text': "So what we're gonna do is we'll start by calculating the entropy and information gain for each of the predictor variable.", 'start': 6289.582, 'duration': 7.261}, {'end': 6301.804, 'text': "So we're gonna start by calculating the information gain and entropy for the road type variable.", 'start': 6297.303, 'duration': 4.501}, {'end': 6305.525, 'text': 'In our data set, you can see that there are four observations.', 'start': 6302.464, 'duration': 3.061}, {'end': 6312.346, 'text': 'There are four observations in the road type column which correspond to the four labels in the speed column.', 'start': 6306.005, 'duration': 6.341}, {'end': 6316.767, 'text': "So we're gonna begin by calculating the information gain of the parent node.", 'start': 6312.944, 'duration': 3.823}, {'end': 6320.089, 'text': 'The parent node is nothing but the speed of the car node.', 'start': 6316.927, 'duration': 3.162}, {'end': 6326.973, 'text': "This is our output variable, correct? It'll be used to show whether the speed of the car is slow or fast.", 'start': 6320.569, 'duration': 6.404}, {'end': 6333.297, 'text': "So to find out the information gain of the speed of the car variable, we'll go through a couple of steps.", 'start': 6327.633, 'duration': 5.664}], 'summary': 'The transcript discusses using information gain and entropy to build a decision tree to classify car speed based on predictor variables like road type, obstruction, and speed limit.', 'duration': 185.632, 'max_score': 6147.665, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk6147665.jpg'}, {'end': 6788.636, 'src': 'embed', 'start': 6758.099, 'weight': 0, 'content': [{'end': 6760.8, 'text': 'Similarly, information gain for speed limit is one.', 'start': 6758.099, 'duration': 2.701}, {'end': 6763.681, 'text': "Now this is the highest value we've got for information gain.", 'start': 6761.2, 'duration': 2.481}, {'end': 6770.443, 'text': "This means that we'll have to use the speed limit variable at our root node in order to split the data set.", 'start': 6764.261, 'duration': 6.182}, {'end': 6773.079, 'text': "right. so guys, don't get confused.", 'start': 6771.056, 'duration': 2.023}, {'end': 6779.706, 'text': 'whichever variable gives you the maximum information gain, that variable has to be chosen at the root node.', 'start': 6773.079, 'duration': 6.627}, {'end': 6782.55, 'text': "so that's why we have the root node as speed limit.", 'start': 6779.706, 'duration': 2.844}, {'end': 6788.636, 'text': "So if you've maintained the speed limit, then you're going to go slow, but if you haven't maintained the speed limit,", 'start': 6783.132, 'duration': 5.504}], 'summary': 'Speed limit has the highest information gain, chosen as root node for data set splitting.', 'duration': 30.537, 'max_score': 6758.099, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk6758099.jpg'}], 'start': 4576.847, 'title': 'Regression and classification algorithms', 'summary': 'Covers data splicing, linear regression model building in python, evaluation metrics, logistic regression, and decision tree structure, including information gain calculation, with a maximum gain of 1 for car speed prediction.', 'chapters': [{'end': 4717.434, 'start': 4576.847, 'title': 'Linear regression and data splicing', 'summary': 'Discusses the process of data splicing, involving the splitting of a data set into training and testing data, with 80% assigned for training and 20% for testing, to understand the linear relationship between the minimum and maximum temperature, utilizing only two variables for the demonstration of linear regression.', 'duration': 140.587, 'highlights': ['The chapter discusses the process of data splicing, involving the splitting of a data set into training and testing data, with 80% assigned for training and 20% for testing.', 'The demonstration of linear regression involves the understanding of the linear relationship between the minimum and maximum temperature, utilizing only two variables.']}, {'end': 5020.03, 'start': 4718.095, 'title': 'Linear regression model in python', 'summary': 'Introduces the process of building a linear regression model in python, including instantiating the linear regression class, finding the intercept and slope values, making predictions, and evaluating model accuracy, suggesting ways to improve accuracy with more data or parameter tuning.', 'duration': 301.935, 'highlights': ['The linear regression class in Python allows for instantiating the class and calling the fit method with training data, enabling the creation of a regression instance and model building process.', 'The model finds the best values for intercept and slope, resulting in a line that best fits the data, with the intercept around 10.66 and coefficient around 0.92, indicating the significance of input variables.', 'The process involves making predictions with the test dataset, comparing actual and predicted values, and visualizing the comparison through a bar graph, suggesting ways to improve accuracy with more training data.']}, {'end': 5870.394, 'start': 5020.03, 'title': 'Regression and classification algorithms', 'summary': 'Explains linear regression, including the evaluation metrics mean absolute error, mean squared error, and root mean squared error, with calculated values of 3.19, 17.63, and 4.19 respectively. it then delves into logistic regression, emphasizing its use for categorical outcome prediction, and concludes with an introduction to decision tree as a classification algorithm.', 'duration': 850.364, 'highlights': ['The chapter explains linear regression and its evaluation metrics, including mean absolute error, mean squared error, and root mean squared error, with calculated values of 3.19, 17.63, and 4.19 respectively.', 'Logistic regression is introduced as a method for predicting categorical outcomes, and the reasoning behind its use instead of linear regression for such cases is detailed.', 'The concept and application of decision tree as a classification algorithm are explained, with an example illustrating its functioning.']}, {'end': 6138.708, 'start': 5870.514, 'title': 'Decision tree structure & id3 algorithm', 'summary': 'Explains the structure of a decision tree, including root nodes, internal nodes, terminal nodes, and branches, and introduces the id3 algorithm for building a decision tree, which involves selecting the best attribute, assigning it as the decision variable for the root node, building descendant nodes, and assigning classification labels to the leaf nodes.', 'duration': 268.194, 'highlights': ['The ID3 algorithm involves six defined steps for building a decision tree. It includes selecting the best attribute, assigning it as the decision variable for the root node, building descendant nodes, and assigning classification labels to the leaf nodes.', 'The structure of a decision tree includes root nodes, internal nodes representing decision points, terminal nodes representing the final class of the output variable, and branches representing connections between nodes.', 'The root node in a decision tree is assigned to a variable that is significant in predicting the output, and internal nodes represent decision points that eventually lead to the output.']}, {'end': 6853.981, 'start': 6139.388, 'title': 'Decision tree for car speed prediction', 'summary': 'Discusses the process of selecting the best attribute for a decision tree by calculating information gain and entropy, using a use case of classifying car speed based on road type, obstruction, and speed limit, with the speed limit variable providing the maximum information gain of 1.', 'duration': 714.593, 'highlights': ['The speed limit variable provides the maximum information gain of 1, indicating that it should be chosen at the root node to split the data set for classifying car speed.', 'Entropy is used to measure the impurity or uncertainty present in the data, while information gain indicates how much information a variable gives about the final outcome.', 'The entropy for the road type variable is calculated as 0.9, indicating a higher level of uncertainty compared to the speed limit variable.']}], 'duration': 2277.134, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk4576847.jpg', 'highlights': ['The speed limit variable provides the maximum information gain of 1 for car speed prediction', 'The chapter discusses the process of data splicing, involving the splitting of a data set into training and testing data, with 80% assigned for training and 20% for testing', 'The demonstration of linear regression involves the understanding of the linear relationship between the minimum and maximum temperature, utilizing only two variables', 'The concept and application of decision tree as a classification algorithm are explained, with an example illustrating its functioning', 'The ID3 algorithm involves six defined steps for building a decision tree, including selecting the best attribute, assigning it as the decision variable for the root node, building descendant nodes, and assigning classification labels to the leaf nodes']}, {'end': 8066.63, 'segs': [{'end': 6968.304, 'src': 'embed', 'start': 6939.372, 'weight': 6, 'content': [{'end': 6942.493, 'text': 'why do we have to use random forest when we already have decision trees?', 'start': 6939.372, 'duration': 3.121}, {'end': 6945.555, 'text': 'There are three main reasons why random forest is used.', 'start': 6942.513, 'duration': 3.042}, {'end': 6952.958, 'text': 'Now even though decision trees are convenient and easily implemented, they are not as accurate as random forest.', 'start': 6946.275, 'duration': 6.683}, {'end': 6960.861, 'text': "Decision trees work very effectively with the training data, but they're not flexible when it comes to classifying a new sample.", 'start': 6953.538, 'duration': 7.323}, {'end': 6964.902, 'text': 'Now this happens because of something known as overfitting.', 'start': 6962.141, 'duration': 2.761}, {'end': 6968.304, 'text': 'Now overfitting is a problem that is seen with decision trees.', 'start': 6965.422, 'duration': 2.882}], 'summary': 'Random forest is used for improved accuracy and flexibility over decision trees due to overfitting issues.', 'duration': 28.932, 'max_score': 6939.372, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk6939372.jpg'}, {'end': 7214.324, 'src': 'heatmap', 'start': 7028.58, 'weight': 0.754, 'content': [{'end': 7031.241, 'text': "But in random forest, there's something known as bagging.", 'start': 7028.58, 'duration': 2.661}, {'end': 7042.345, 'text': 'Now, the basic idea behind bagging is to reduce the variations in the predictions by combining the result of multiple decision trees on different samples of the data set.', 'start': 7031.701, 'duration': 10.644}, {'end': 7049.427, 'text': "So your data set will be divided into different samples and you'll be building a decision tree on each of these samples.", 'start': 7042.765, 'duration': 6.662}, {'end': 7053.829, 'text': 'This way, each decision tree will be studying one subset of your data.', 'start': 7049.888, 'duration': 3.941}, {'end': 7060.233, 'text': 'So this way, overfitting will get reduced because one decision tree is not studying the entire data set.', 'start': 7054.369, 'duration': 5.864}, {'end': 7062.394, 'text': "Now let's focus on the random forest.", 'start': 7060.693, 'duration': 1.701}, {'end': 7065.957, 'text': 'Now in order to understand random forest, we look at a small example.', 'start': 7062.975, 'duration': 2.982}, {'end': 7068.078, 'text': 'Consider this data set.', 'start': 7066.757, 'duration': 1.321}, {'end': 7070.72, 'text': 'In this data, we have four predictor variables.', 'start': 7068.298, 'duration': 2.422}, {'end': 7074.862, 'text': 'We have blood flow, blocked arteries, chest pain, and weight.', 'start': 7071.06, 'duration': 3.802}, {'end': 7080.226, 'text': 'Now these variables are used to predict whether or not a person has a heart disease.', 'start': 7075.543, 'duration': 4.683}, {'end': 7086.838, 'text': "So we're going to use this dataset to create a random forest that predicts if a person has a heart disease or not.", 'start': 7080.876, 'duration': 5.962}, {'end': 7092.52, 'text': 'Now the first step in creating a random forest is that you create a bootstrapped dataset.', 'start': 7087.418, 'duration': 5.102}, {'end': 7098.842, 'text': 'Now in bootstrapping, all you have to do is you have to randomly select samples from your original dataset.', 'start': 7093.08, 'duration': 5.762}, {'end': 7104.384, 'text': 'Okay, and a point to note is that you can select the same sample more than once.', 'start': 7099.482, 'duration': 4.902}, {'end': 7111.405, 'text': 'So if you look at the original data set, we have abnormal, normal, normal, and abnormal.', 'start': 7105.761, 'duration': 5.644}, {'end': 7113.046, 'text': 'Look at the blood flow section.', 'start': 7111.485, 'duration': 1.561}, {'end': 7119.211, 'text': "Now here I've randomly selected samples, normal, abnormal, and I've selected one sample twice.", 'start': 7113.787, 'duration': 5.424}, {'end': 7122.074, 'text': 'You can do this in a bootstrap data set.', 'start': 7120.092, 'duration': 1.982}, {'end': 7125.737, 'text': 'Now all I did here is I created a bootstrap data set.', 'start': 7122.614, 'duration': 3.123}, {'end': 7132.402, 'text': 'Bootstrapping is nothing but an estimation method used to make predictions on a data by resampling the data.', 'start': 7126.197, 'duration': 6.205}, {'end': 7134.504, 'text': 'This is a bootstrap data set.', 'start': 7133.023, 'duration': 1.481}, {'end': 7140.665, 'text': "Now even though this seems very simple, in real world problems you'll never get such a small data set.", 'start': 7134.984, 'duration': 5.681}, {'end': 7144.606, 'text': 'Okay, so bootstrapping is actually a little more complex than this.', 'start': 7141.145, 'duration': 3.461}, {'end': 7151.987, 'text': 'Usually in real world problems you have a huge data set and bootstrapping that data set is actually a pretty complex problem.', 'start': 7145.206, 'duration': 6.781}, {'end': 7157.228, 'text': "Now here because I'm making you understand how random forest works, so that's why I've considered a small data set.", 'start': 7152.307, 'duration': 4.921}, {'end': 7162.889, 'text': "Now you're going to use the bootstrap data set that you created and you're going to build decision trees from it.", 'start': 7157.988, 'duration': 4.901}, {'end': 7168.672, 'text': 'Now one more thing to note in random forest is you will not be using your entire data set.', 'start': 7163.63, 'duration': 5.042}, {'end': 7172.053, 'text': "So you'll only be using few of the variables at each node.", 'start': 7169.172, 'duration': 2.881}, {'end': 7176.674, 'text': "So for example, we'll only consider two variables at each step.", 'start': 7172.513, 'duration': 4.161}, {'end': 7182.556, 'text': 'So if we begin at the root node, here we will randomly select two variables as candidates for the root node.', 'start': 7177.095, 'duration': 5.461}, {'end': 7186.938, 'text': "Let's say that we selected blood flow and blocked arteries.", 'start': 7183.437, 'duration': 3.501}, {'end': 7192.179, 'text': 'Out of these two variables, we have to select the variable that best separates the sample.', 'start': 7187.798, 'duration': 4.381}, {'end': 7200.141, 'text': "So for the sake of this example, let's say that blocked arteries is the most significant predictor and that's why we'll assign it to the root node.", 'start': 7193.059, 'duration': 7.082}, {'end': 7205.222, 'text': 'Now our next step is to repeat the same process for each of these upcoming branch nodes.', 'start': 7200.881, 'duration': 4.341}, {'end': 7214.324, 'text': "Here we'll again select two variables at random as candidates for each of these branch nodes and then choose a variable that best separates the samples.", 'start': 7205.682, 'duration': 8.642}], 'summary': 'Random forest uses bagging to reduce overfitting by combining multiple decision trees on different samples of the dataset.', 'duration': 185.744, 'max_score': 7028.58, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7028580.jpg'}, {'end': 7244.602, 'src': 'embed', 'start': 7222.632, 'weight': 9, 'content': [{'end': 7230.935, 'text': "you'll randomly select a couple of variables for each node and then you'll calculate which variable best splits the data at that node.", 'start': 7222.632, 'duration': 8.303}, {'end': 7240.659, 'text': "So for each node we'll randomly select two or three variables and out of those two, three variables we'll see which variable best separates the data.", 'start': 7231.275, 'duration': 9.384}, {'end': 7244.602, 'text': 'Okay, so at each node will be calculating information gain and entropy.', 'start': 7241.139, 'duration': 3.463}], 'summary': 'Randomly select variables for each node to calculate best split; calculate information gain and entropy.', 'duration': 21.97, 'max_score': 7222.632, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7222632.jpg'}, {'end': 7345.733, 'src': 'embed', 'start': 7319.124, 'weight': 0, 'content': [{'end': 7326.727, 'text': 'So having a variety of decision trees in a random forest is what makes it more effective than an individual decision tree.', 'start': 7319.124, 'duration': 7.603}, {'end': 7333.051, 'text': 'So, instead of having an individual decision tree, which was created using all the features,', 'start': 7328.03, 'duration': 5.021}, {'end': 7341.373, 'text': 'you can build a random forest that uses multiple decision trees, wherein each decision tree has a random set of predictor variables.', 'start': 7333.051, 'duration': 8.322}, {'end': 7345.733, 'text': 'Now step number four is predicting the outcome of a new data point.', 'start': 7342.253, 'duration': 3.48}], 'summary': 'Random forest uses multiple decision trees with random predictor variables, making it more effective than an individual decision tree.', 'duration': 26.609, 'max_score': 7319.124, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7319124.jpg'}, {'end': 7429.058, 'src': 'embed', 'start': 7403.641, 'weight': 4, 'content': [{'end': 7410.824, 'text': "Okay, let's say that three decision trees said that yes, the patient has heart disease and one decision tree said that no, it doesn't have.", 'start': 7403.641, 'duration': 7.183}, {'end': 7417.794, 'text': 'So this means you will obviously classify the patient as having a heart disease because three of them voted for yes.', 'start': 7411.391, 'duration': 6.403}, {'end': 7419.834, 'text': "It's based on majority.", 'start': 7418.394, 'duration': 1.44}, {'end': 7425.256, 'text': 'So guys I hope the concept behind random forest is understandable.', 'start': 7420.995, 'duration': 4.261}, {'end': 7429.058, 'text': 'Now the next step is you will evaluate the efficiency of the model.', 'start': 7425.717, 'duration': 3.341}], 'summary': 'Three decision trees predict heart disease; majority voting determines patient classification. next step: evaluate model efficiency.', 'duration': 25.417, 'max_score': 7403.641, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7403641.jpg'}, {'end': 7531.717, 'src': 'embed', 'start': 7509.528, 'weight': 2, 'content': [{'end': 7518.252, 'text': 'So eventually you can measure the accuracy of a random forest by the proportion of out of bag samples that are correctly classified.', 'start': 7509.528, 'duration': 8.724}, {'end': 7523.674, 'text': 'Right, because the out of bag data set is used to evaluate the efficiency of your model.', 'start': 7518.792, 'duration': 4.882}, {'end': 7531.717, 'text': 'So you can calculate the accuracy by understanding how many samples was this out of bag data set correctly able to classify.', 'start': 7523.994, 'duration': 7.723}], 'summary': 'Random forest accuracy measured by proportion of correctly classified out of bag samples.', 'duration': 22.189, 'max_score': 7509.528, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7509528.jpg'}, {'end': 7578.014, 'src': 'embed', 'start': 7552.859, 'weight': 8, 'content': [{'end': 7558.642, 'text': 'A bootstrap data set is nothing but randomly selected observations from your original data set,', 'start': 7552.859, 'duration': 5.783}, {'end': 7562.124, 'text': 'and you can also have duplicate values in your bootstrap data set.', 'start': 7558.642, 'duration': 3.482}, {'end': 7571.03, 'text': "The next step is you're going to create a decision tree by considering a random set of predictor variables for each decision tree.", 'start': 7563.285, 'duration': 7.745}, {'end': 7578.014, 'text': "So the third step is you'll go back to step one, create a bootstrap data set, again create a decision tree.", 'start': 7571.45, 'duration': 6.564}], 'summary': 'Bootstrap data set with duplicates used to create decision trees.', 'duration': 25.155, 'max_score': 7552.859, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7552859.jpg'}, {'end': 7682.923, 'src': 'embed', 'start': 7654.835, 'weight': 10, 'content': [{'end': 7661.043, 'text': 'Naive bias is again a supervised classification algorithm which is based on the Bayes theorem.', 'start': 7654.835, 'duration': 6.208}, {'end': 7665.275, 'text': 'Now the Bayes theorem basically follows a probabilistic approach.', 'start': 7661.974, 'duration': 3.301}, {'end': 7674.819, 'text': 'The main idea behind naive bias is that the predictor variables in a machine learning model are independent of each other,', 'start': 7665.816, 'duration': 9.003}, {'end': 7682.923, 'text': 'meaning that the outcome of a model depends on a set of independent variables that have nothing to do with each other.', 'start': 7674.819, 'duration': 8.104}], 'summary': 'Naive bayes is a supervised classification algorithm based on bayes theorem and assumes predictor variables are independent.', 'duration': 28.088, 'max_score': 7654.835, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7654835.jpg'}, {'end': 7792.49, 'src': 'embed', 'start': 7765.652, 'weight': 11, 'content': [{'end': 7769.814, 'text': 'Now consider this data set of around 1, 500 observations.', 'start': 7765.652, 'duration': 4.162}, {'end': 7773.337, 'text': 'Okay, here we have the following output classes.', 'start': 7770.395, 'duration': 2.942}, {'end': 7776.179, 'text': 'We have either cat, parrot, or turtle.', 'start': 7773.777, 'duration': 2.402}, {'end': 7784.044, 'text': 'Okay, these are our output classes, and the predictive variables are swim, wings, green color, and sharp teeth.', 'start': 7776.519, 'duration': 7.525}, {'end': 7792.49, 'text': 'So basically your type is your output variable and swim, wings, green, and sharp teeth are your predictor variables.', 'start': 7785.128, 'duration': 7.362}], 'summary': 'Dataset has 1,500 observations with output classes: cat, parrot, or turtle, and predictive variables: swim, wings, green color, and sharp teeth.', 'duration': 26.838, 'max_score': 7765.652, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk7765652.jpg'}, {'end': 8040.633, 'src': 'embed', 'start': 8010.055, 'weight': 3, 'content': [{'end': 8010.775, 'text': 'all right.', 'start': 8010.055, 'duration': 0.72}, {'end': 8012.617, 'text': 'so, guys, this is how naive bias works.', 'start': 8010.775, 'duration': 1.842}, {'end': 8016.299, 'text': 'you basically calculate the conditional probability at each step.', 'start': 8012.617, 'duration': 3.682}, {'end': 8021.207, 'text': 'whatever classification needs to be done, that has to be calculated through probability.', 'start': 8016.926, 'duration': 4.281}, {'end': 8029.19, 'text': "There's a lot of statistic that comes into naive bias, and if you all want to learn more about statistics and probability,", 'start': 8021.707, 'duration': 7.483}, {'end': 8030.55, 'text': "I'll leave a link in the description.", 'start': 8029.19, 'duration': 1.36}, {'end': 8032.27, 'text': 'You all can watch that video as well.', 'start': 8030.89, 'duration': 1.38}, {'end': 8038.532, 'text': "There I've explained exactly what conditional probability is and the bias theorem is also explained very well.", 'start': 8032.671, 'duration': 5.861}, {'end': 8040.633, 'text': 'So you all can check out that video also.', 'start': 8038.552, 'duration': 2.081}], 'summary': 'Naive bias works by calculating conditional probability for classification, with emphasis on statistics and probability.', 'duration': 30.578, 'max_score': 8010.055, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8010055.jpg'}], 'start': 6854.637, 'title': 'Random forest and naive bias', 'summary': 'Covers decision trees, random forest basics, and advantages, explaining bootstrapping, model efficiency evaluation using out-of-bag data set, and introduces naive bias classification, with a demonstration predicting animal type with a probability of 0.066.', 'chapters': [{'end': 7125.737, 'start': 6854.637, 'title': 'Decision trees and random forest', 'summary': 'Covers the basics of decision trees, including entropy calculation, information gain, and overfitting issues, as well as the advantages of random forest such as reducing overfitting through bagging and building multiple decision trees for more accurate predictions.', 'duration': 271.1, 'highlights': ['Random forest basically builds multiple decision trees and glues them together to get a more accurate and stable prediction.', 'Overfitting occurs when a model studies the training data to such an extent that it negatively influences the performance of the model on the new data.', 'Random forest reduces overfitting by combining the result of multiple decision trees on different samples of the data set through bagging.', 'Decision trees work effectively with the training data but are not flexible when classifying new samples due to overfitting.', 'Random forest is used for more accurate predictions compared to decision trees and is advantageous in reducing overfitting through bagging and building multiple decision trees.']}, {'end': 7425.256, 'start': 7126.197, 'title': 'Random forest overview', 'summary': 'Introduces the concept of bootstrapping and explains how random forest works, utilizing multiple decision trees with random subsets of predictor variables to predict outcomes for new data points.', 'duration': 299.059, 'highlights': ['Random forest involves building multiple decision trees with a random set of predictor variables.', 'The process of bootstrapping is used to create a bootstrap data set and build decision trees from it.', 'Prediction in random forest is based on the majority votes from the decision trees.', 'Information gain and entropy are calculated at each node to select the best variable for splitting the data.']}, {'end': 7765.131, 'start': 7425.717, 'title': 'Random forest & naive bias', 'summary': 'Explains the process of evaluating the efficiency of a model using out-of-bag data set in random forest, where one third of the original data set is not included in the bootstrap data set and is used to check the accuracy of the model. it also introduces naive bias as a supervised classification algorithm based on the bayes theorem and explains the principle behind it.', 'duration': 339.414, 'highlights': ['The out-of-bag data set, comprising about one third of the original data set not included in the bootstrap data set, is used to evaluate the efficiency of the random forest model by measuring the accuracy of classification.', 'A bootstrap data set is created by randomly selecting observations from the original data set, allowing duplicate values, and a decision tree is created by considering a random set of predictor variables for each decision tree.', "Naive bias is introduced as a supervised classification algorithm based on the Bayes theorem, which assumes that predictor variables in a machine learning model are independent of each other, and it is named 'naive' due to this assumption."]}, {'end': 8066.63, 'start': 7765.652, 'title': 'Naive bias classification', 'summary': 'Discusses a dataset of 1500 observations and three output classes: cat, parrot, and turtle, with predictor variables swim, wings, green color, and sharp teeth. by calculating conditional probabilities, it predicts the animal type based on given parameters, ultimately concluding that the animal is a turtle with a probability of 0.066, demonstrating how naive bias classification works.', 'duration': 300.978, 'highlights': ['The dataset consists of 1500 observations with output classes: cat, parrot, and turtle, and predictor variables swim, wings, green color, and sharp teeth.', 'Out of 500 cats, 450 can swim (90%), 0 have wings, 0 are green, and 500 have sharp teeth.', 'Out of 500 parrots, 50 can swim (10%), 500 have wings, 400 are green, and 0 have sharp teeth.', 'All 500 turtles can swim, 0 have wings, 100 are green (20%), and 50 have sharp teeth.', 'The goal is to predict whether the animal is a cat, parrot, or turtle based on the given parameters of swim, wings, green, and sharp teeth.', 'The calculation of conditional probabilities leads to the conclusion that the animal is a turtle with a probability of 0.066.', 'Naive bias classification involves calculating conditional probabilities at each step to make predictions based on probabilities of given parameters.']}], 'duration': 1211.993, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk6854637.jpg', 'highlights': ['Random forest basically builds multiple decision trees for more accurate predictions.', 'Random forest reduces overfitting by combining results of multiple decision trees.', 'Out-of-bag data set is used to evaluate the efficiency of the random forest model.', 'Naive bias classification involves calculating conditional probabilities for predictions.', 'Prediction in random forest is based on majority votes from decision trees.', 'Random forest is advantageous in reducing overfitting through bagging and building multiple decision trees.', 'Decision trees work effectively with training data but are not flexible when classifying new samples.', 'Random forest involves building multiple decision trees with a random set of predictor variables.', 'The process of bootstrapping is used to create a bootstrap data set and build decision trees from it.', 'Information gain and entropy are calculated at each node to select the best variable for splitting the data.', 'Naive bias is introduced as a supervised classification algorithm based on the Bayes theorem.', 'The dataset consists of 1500 observations with output classes: cat, parrot, and turtle.']}, {'end': 9858.759, 'segs': [{'end': 8239.168, 'src': 'embed', 'start': 8153.892, 'weight': 0, 'content': [{'end': 8161.116, 'text': 'In this manner, the KNN algorithm classifies the data points based on how similar they are to their neighboring data points right?', 'start': 8153.892, 'duration': 7.224}, {'end': 8162.816, 'text': 'So this is a small example.', 'start': 8161.616, 'duration': 1.2}, {'end': 8165.818, 'text': "We'll discuss more about it in the further slides.", 'start': 8162.856, 'duration': 2.962}, {'end': 8170.019, 'text': 'Now let me tell you a couple of features of KNN algorithm.', 'start': 8166.733, 'duration': 3.286}, {'end': 8173.245, 'text': 'First of all, we know that it is a supervised learning algorithm.', 'start': 8170.38, 'duration': 2.865}, {'end': 8177.192, 'text': 'It uses labeled input data set to predict the output of the data points.', 'start': 8173.525, 'duration': 3.667}, {'end': 8185.215, 'text': 'then it is also one of the simplest machine learning algorithms and it can be easily implemented for a varied set of problems.', 'start': 8178.032, 'duration': 7.183}, {'end': 8191.277, 'text': 'Another feature is that it is non-parametric, meaning that it does not take in any assumptions.', 'start': 8185.895, 'duration': 5.382}, {'end': 8199.741, 'text': 'For example, naive bias is a parametric model because it assumes that all the independent variables are in no way related to each other.', 'start': 8191.818, 'duration': 7.923}, {'end': 8202.361, 'text': 'It has assumptions about the model.', 'start': 8200.281, 'duration': 2.08}, {'end': 8207.304, 'text': "K nearest neighbor has no such assumptions, that's why it's considered a non-parametric model.", 'start': 8202.722, 'duration': 4.582}, {'end': 8210.946, 'text': 'Another feature is that it is a lazy algorithm.', 'start': 8207.964, 'duration': 2.982}, {'end': 8221.153, 'text': 'Now lazy algorithm basically is any algorithm that memorizes the training set instead of learning a discriminative function from the training data.', 'start': 8211.826, 'duration': 9.327}, {'end': 8228.197, 'text': 'Now even though KNN is mainly a classification algorithm, it can also be used for regression cases.', 'start': 8222.452, 'duration': 5.745}, {'end': 8233.981, 'text': 'So KNN is actually both a classification and a regression algorithm.', 'start': 8229.477, 'duration': 4.504}, {'end': 8239.168, 'text': "but mostly you'll see that it'll be used only for classification problems.", 'start': 8234.584, 'duration': 4.584}], 'summary': 'Knn is a supervised learning algorithm, non-parametric, and can be used for both classification and regression.', 'duration': 85.276, 'max_score': 8153.892, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8153892.jpg'}, {'end': 8286.541, 'src': 'embed', 'start': 8259.258, 'weight': 5, 'content': [{'end': 8266.764, 'text': 'Now the problem statement is to assign the new input data point to one of the two classes by using the KNN algorithm.', 'start': 8259.258, 'duration': 7.506}, {'end': 8272.035, 'text': 'So the first step in the KNN algorithm is to define the value of K.', 'start': 8267.272, 'duration': 4.763}, {'end': 8275.236, 'text': 'But what does the K in the KNN algorithm stand for?', 'start': 8272.035, 'duration': 3.201}, {'end': 8281.539, 'text': "Now, the K stands for the number of nearest neighbors, and that's why it's got the name K nearest neighbors.", 'start': 8275.656, 'duration': 5.883}, {'end': 8286.541, 'text': "Now in this image I've defined the value of K as three.", 'start': 8282.54, 'duration': 4.001}], 'summary': 'Using knn algorithm to assign input data to classes, k defines number of nearest neighbors, k=3.', 'duration': 27.283, 'max_score': 8259.258, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8259258.jpg'}, {'end': 8511.399, 'src': 'embed', 'start': 8485.094, 'weight': 7, 'content': [{'end': 8491.196, 'text': 'Even though SVM is mainly used for classification, there is something known as the support vector regressor right?', 'start': 8485.094, 'duration': 6.102}, {'end': 8492.957, 'text': 'That is used for regression problems.', 'start': 8491.276, 'duration': 1.681}, {'end': 8499.12, 'text': 'Now SVM can also be used to classify nonlinear data by using kernel tricks.', 'start': 8493.698, 'duration': 5.422}, {'end': 8504.474, 'text': 'Nonlinear data is basically data that cannot be separated by using a single linear line.', 'start': 8499.77, 'duration': 4.704}, {'end': 8507.176, 'text': "I'll be talking more about this in the upcoming slides.", 'start': 8504.934, 'duration': 2.242}, {'end': 8511.399, 'text': "Now let's move on and discuss how SVM works.", 'start': 8508.117, 'duration': 3.282}], 'summary': 'Svm can be used for regression and nonlinear classification using kernel tricks.', 'duration': 26.305, 'max_score': 8485.094, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8485094.jpg'}, {'end': 8580.163, 'src': 'embed', 'start': 8551.997, 'weight': 6, 'content': [{'end': 8560.019, 'text': 'It draws a decision boundary which is nothing but a hyperplane between any two classes in order to separate them or classify them.', 'start': 8551.997, 'duration': 8.022}, {'end': 8563.98, 'text': "Now, I know that you're thinking how do you know where to draw a hyperplane?", 'start': 8560.459, 'duration': 3.521}, {'end': 8570.301, 'text': 'Now, the basic principle behind SVM is to draw a hyperplane that best separates the two classes.', 'start': 8564.44, 'duration': 5.861}, {'end': 8573.462, 'text': 'In our case, the two classes are the rabbits and the wolves.', 'start': 8570.941, 'duration': 2.521}, {'end': 8580.163, 'text': "All right, now before we move any further, let's discuss the different terminologies that are there in support vector machines.", 'start': 8573.922, 'duration': 6.241}], 'summary': 'Svm uses hyperplanes to separate classes, like rabbits and wolves.', 'duration': 28.166, 'max_score': 8551.997, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8551997.jpg'}, {'end': 9368.447, 'src': 'embed', 'start': 9344.131, 'weight': 10, 'content': [{'end': 9349.856, 'text': "So here I'm calculating the accuracy on the training data set and on the testing data set.", 'start': 9344.131, 'duration': 5.725}, {'end': 9353.079, 'text': "So let's look at the output of this.", 'start': 9351.177, 'duration': 1.902}, {'end': 9355.921, 'text': 'Now guys, ignore this future warning.', 'start': 9353.899, 'duration': 2.022}, {'end': 9359.965, 'text': 'Warnings are ignored in Python.', 'start': 9355.941, 'duration': 4.024}, {'end': 9368.447, 'text': 'Now, accuracy of the logistic regression classifier on the training data set is around 70%, which is pretty good on the training data set,', 'start': 9360.603, 'duration': 7.844}], 'summary': 'Logistic regression classifier achieves 70% accuracy on training data.', 'duration': 24.316, 'max_score': 9344.131, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9344131.jpg'}, {'end': 9418.291, 'src': 'embed', 'start': 9392.162, 'weight': 9, 'content': [{'end': 9396.728, 'text': "We'll calculate the accuracy of the decision tree on the training and the testing data set.", 'start': 9392.162, 'duration': 4.566}, {'end': 9401.633, 'text': 'So if you do that for a decision tree on the training data set, you get 100% accuracy.', 'start': 9397.188, 'duration': 4.445}, {'end': 9408.565, 'text': 'but on the testing data set, you have around 87% of accuracy.', 'start': 9404.082, 'duration': 4.483}, {'end': 9418.291, 'text': 'Now this is something that I discussed with you all earlier that decision trees are very good with training data set because of a process known as overfitting.', 'start': 9409.065, 'duration': 9.226}], 'summary': 'Decision tree achieved 100% accuracy on training data, but 87% on testing data, indicating overfitting.', 'duration': 26.129, 'max_score': 9392.162, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9392162.jpg'}, {'end': 9465.108, 'src': 'embed', 'start': 9434.27, 'weight': 8, 'content': [{'end': 9436.07, 'text': 'Coming to KNN classifier again.', 'start': 9434.27, 'duration': 1.8}, {'end': 9444.493, 'text': 'all you have to do is you have to call the KNneighbor classifier this function and you have to fit this with the training data set.', 'start': 9436.07, 'duration': 8.423}, {'end': 9451.078, 'text': 'If you calculate the accuracy for a KNN classifier, we get a good accuracy actually.', 'start': 9445.151, 'duration': 5.927}, {'end': 9458.584, 'text': 'On the training data set, we get an accuracy of 95% and on the testing data set is 100%.', 'start': 9451.498, 'duration': 7.086}, {'end': 9465.108, 'text': 'That is really good because our testing data set actually achieved more of an accuracy than on our training data set.', 'start': 9458.584, 'duration': 6.524}], 'summary': 'Using knn classifier, achieved 95% accuracy on training data and 100% accuracy on testing data.', 'duration': 30.838, 'max_score': 9434.27, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9434270.jpg'}, {'end': 9847.794, 'src': 'embed', 'start': 9816.346, 'weight': 11, 'content': [{'end': 9819.791, 'text': 'Group three is 36 and 39 or something like that.', 'start': 9816.346, 'duration': 3.445}, {'end': 9824.116, 'text': "So let's say you're trying to cluster people into different groups based on their age.", 'start': 9820.211, 'duration': 3.905}, {'end': 9828.32, 'text': 'So for such problems, you can make use of the key means clustering algorithm.', 'start': 9824.737, 'duration': 3.583}, {'end': 9833.304, 'text': 'One of the major applications of the clustering algorithm is seen in targeted marketing.', 'start': 9828.8, 'duration': 4.504}, {'end': 9836.166, 'text': "I don't know how many of you are aware of targeted marketing.", 'start': 9833.765, 'duration': 2.401}, {'end': 9842.572, 'text': 'Targeted marketing is all about marketing a specific product to a specific audience.', 'start': 9836.607, 'duration': 5.965}, {'end': 9847.794, 'text': "Let's say you're trying to sell fancy clothes or a fancy set of bags and all of that.", 'start': 9843.212, 'duration': 4.582}], 'summary': 'K-means clustering can group 36 and 39-year-olds for targeted marketing.', 'duration': 31.448, 'max_score': 9816.346, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9816346.jpg'}], 'start': 8067.45, 'title': 'Introduction to k-nearest neighbor algorithm and other classification techniques', 'summary': 'Introduces the k-nearest neighbor (knn) algorithm, explaining its features and usage of labeled input data, non-parametric nature, and determining the value of k. it also discusses support vector machine (svm) algorithm, implementation and accuracy of various classification algorithms including logistic regression, decision tree, knn, naive bias, and svm, concluding with an introduction to k-means clustering and its applications in targeted marketing.', 'chapters': [{'end': 8370.458, 'start': 8067.45, 'title': 'Understanding k-nearest neighbor algorithm', 'summary': 'Introduces the k-nearest neighbor (knn) algorithm, a supervised classification algorithm, and explains its key features, such as using labeled input data, non-parametric nature, and the process of determining the value of k for classifying new data points.', 'duration': 303.008, 'highlights': ['K-nearest neighbor algorithm is a supervised classification algorithm that classifies new data points into the target class based on the features of its neighboring data points', 'Supervised learning algorithm uses labeled input data set to predict the output of the data points', 'KNN is non-parametric, meaning that it does not make any assumptions about the underlying data distribution', 'KNN is a lazy algorithm that memorizes the training set instead of learning a discriminative function from the training data', 'KNN can be used for both classification and regression problems, although it is primarily used for classification', 'The value of K in the KNN algorithm represents the number of nearest neighbors considered to classify a new data point', 'The process of determining the suitable value of K for KNN is discussed, mentioning methods like the ELBO method']}, {'end': 9300.206, 'start': 8370.458, 'title': 'Knn and svm algorithms', 'summary': 'Discusses the simple math behind the k nearest neighbor (knn) algorithm and explains the usage of euclidean distance. it then delves into the support vector machine (svm) algorithm, highlighting its use of hyperplanes and kernel tricks to classify data, followed by a practical demonstration of implementing multiple classification algorithms using scikit-learn.', 'duration': 929.748, 'highlights': ['KNN uses the Euclidean distance to check the closeness of a new data point with its neighbors, making it a commonly used algorithm.', 'SVM is a supervised learning algorithm that uses hyperplanes as decision boundaries between classes and can be employed for both regression and classification tasks.', 'SVM can be used to classify nonlinear data by using kernel tricks, transforming data into higher dimensions to create clear dividing margins between classes.', 'A practical demonstration of implementing multiple classification algorithms using scikit-learn is conducted, emphasizing the importance of normalizing data for unbiased outcomes.']}, {'end': 9858.759, 'start': 9301.067, 'title': 'Classification algorithms and k-means clustering', 'summary': 'Discussed the implementation and accuracy of various classification algorithms including logistic regression (70% on training, 40% on testing), decision tree (100% on training, 87% on testing), knn (95% on training, 100% on testing), naive bias (86% on training, 67% on testing), and support vector machines (61% on training, 33% on testing). it also explained the confusion matrix and its measures, and concluded with an introduction to k-means clustering and its applications in targeted marketing.', 'duration': 557.692, 'highlights': ['The KNN classifier achieved the highest accuracy of 100% on the testing data set, outperforming other classifiers.', 'The decision tree classifier showed 100% accuracy on the training data set and 87% accuracy on the testing data set, making it the second most accurate classifier.', 'The logistic regression classifier displayed 70% accuracy on the training data set and 40% accuracy on the testing data set, indicating its lower performance compared to other classifiers.', 'Introduction to k-means clustering and its application in targeted marketing, emphasizing its role in grouping similar elements or data points into clusters for effective audience targeting.']}], 'duration': 1791.309, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk8067450.jpg', 'highlights': ["K-nearest neighbor algorithm classifies new data points based on neighboring data points' features", 'Supervised learning algorithm uses labeled input data to predict output', 'KNN is non-parametric, making no assumptions about underlying data distribution', 'KNN is a lazy algorithm that memorizes training set instead of learning a discriminative function', 'KNN can be used for both classification and regression problems', 'Value of K in KNN represents number of nearest neighbors considered for classification', 'SVM uses hyperplanes as decision boundaries between classes for regression and classification tasks', 'SVM can classify nonlinear data using kernel tricks to create clear dividing margins', 'KNN classifier achieved highest accuracy of 100% on testing data set', 'Decision tree classifier showed 100% accuracy on training data set and 87% accuracy on testing data set', 'Logistic regression classifier displayed 70% accuracy on training data set and 40% accuracy on testing data set', 'Introduction to k-means clustering and its application in targeted marketing']}, {'end': 11488.283, 'segs': [{'end': 9892.644, 'src': 'embed', 'start': 9859.219, 'weight': 0, 'content': [{'end': 9863.7, 'text': 'your product is marketed to a specific audience that might be interested in it.', 'start': 9859.219, 'duration': 4.481}, {'end': 9865.861, 'text': 'That is what targeted marketing is.', 'start': 9864.14, 'duration': 1.721}, {'end': 9869.701, 'text': 'So k-means clustering is used majorly in targeted marketing.', 'start': 9866.241, 'duration': 3.46}, {'end': 9878.003, 'text': 'A lot of e-commerce websites like Amazon, Flipkart, eBay, all of these make use of clustering algorithms in order to target the right audience.', 'start': 9870.142, 'duration': 7.861}, {'end': 9880.604, 'text': "Now let's see how the k-means clustering works.", 'start': 9878.483, 'duration': 2.121}, {'end': 9884.478, 'text': 'Now the k in k-means denotes the number of clusters.', 'start': 9881.275, 'duration': 3.203}, {'end': 9892.644, 'text': "Let's say I give you a data set containing 20 points and you want to cluster this data set into four clusters.", 'start': 9885.118, 'duration': 7.526}], 'summary': 'K-means clustering is used in targeted marketing by e-commerce websites like amazon, flipkart, and ebay to cluster data into specific audience segments.', 'duration': 33.425, 'max_score': 9859.219, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9859219.jpg'}, {'end': 10155.376, 'src': 'embed', 'start': 10126.718, 'weight': 3, 'content': [{'end': 10131.861, 'text': 'You start off by computing the sum of squared errors for some values of k.', 'start': 10126.718, 'duration': 5.143}, {'end': 10140.646, 'text': 'Now sum of squared error is basically the sum of the squared distance between each member of the cluster and its centroid.', 'start': 10132.541, 'duration': 8.105}, {'end': 10145.83, 'text': 'So you basically calculate the sum of squared errors for different values of k.', 'start': 10141.067, 'duration': 4.763}, {'end': 10150.772, 'text': 'For example, you can consider k value as two, four, six, eight, 10, 12.', 'start': 10145.83, 'duration': 4.942}, {'end': 10155.376, 'text': 'Consider all these values, compute the sum of squared errors for each of these values.', 'start': 10150.773, 'duration': 4.603}], 'summary': 'Compute sum of squared errors for k values 2, 4, 6, 8, 10, 12.', 'duration': 28.658, 'max_score': 10126.718, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk10126718.jpg'}, {'end': 10588.348, 'src': 'embed', 'start': 10560.445, 'weight': 2, 'content': [{'end': 10563.928, 'text': 'There are actually better ways you can compress information and image.', 'start': 10560.445, 'duration': 3.483}, {'end': 10569.153, 'text': 'So basically I showed you this example because I want you to understand the power of k-means algorithm.', 'start': 10564.368, 'duration': 4.785}, {'end': 10573.697, 'text': 'You can cluster a data set that is this huge into just 16 colors.', 'start': 10569.493, 'duration': 4.204}, {'end': 10576.94, 'text': 'Initially there were 16 million and now you can cluster it to 16 colors.', 'start': 10574.097, 'duration': 2.843}, {'end': 10584.065, 'text': 'So guys, k-means plays a very huge role in computer vision, image processing, object detection, and so on.', 'start': 10578.221, 'duration': 5.844}, {'end': 10588.348, 'text': "It's a very important algorithm when it comes to detecting objects.", 'start': 10584.566, 'duration': 3.782}], 'summary': 'K-means algorithm can compress 16m colors into 16, crucial in image processing and object detection.', 'duration': 27.903, 'max_score': 10560.445, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk10560445.jpg'}, {'end': 10651.836, 'src': 'embed', 'start': 10616.928, 'weight': 1, 'content': [{'end': 10626.094, 'text': "I mean that it's been used in applications such as self-driving cars and is also a part of a lot of deep learning applications such as AlphaGo and so on.", 'start': 10616.928, 'duration': 9.166}, {'end': 10630.197, 'text': 'So reinforcement learning has a different concept to it itself.', 'start': 10626.614, 'duration': 3.583}, {'end': 10632.619, 'text': "So we'll be discussing all the concepts under it.", 'start': 10630.498, 'duration': 2.121}, {'end': 10637.163, 'text': 'So just to brush up your information about reinforcement learning.', 'start': 10633.16, 'duration': 4.003}, {'end': 10651.836, 'text': 'Reinforcement. learning is a part of machine learning where an agent is put in an unknown environment and he learns how to behave in this environment by performing certain actions and observing the rewards which it gets from these actions.', 'start': 10637.504, 'duration': 14.332}], 'summary': 'Reinforcement learning is used in self-driving cars and deep learning apps like alphago. it involves an agent learning to behave in an unknown environment by observing rewards from actions.', 'duration': 34.908, 'max_score': 10616.928, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk10616928.jpg'}], 'start': 9859.219, 'title': 'K-means clustering and reinforcement learning basics', 'summary': 'Explores the application of k-means clustering in targeted marketing and image compression, demonstrating its effectiveness using examples from e-commerce websites and image compression with 16 million colors. additionally, it introduces the basics and concepts of reinforcement learning, highlighting its role in machine learning, applications in self-driving cars, and key components including the agent, environment, state, action, reward, policy, value, and action value, as well as the q-learning algorithm.', 'chapters': [{'end': 10280.263, 'start': 9859.219, 'title': 'K-means clustering in targeted marketing', 'summary': 'Explains how k-means clustering is used in targeted marketing, with examples from e-commerce websites like amazon, flipkart, and ebay, as well as the process of k-means clustering and the elbo method for finding the optimal k value.', 'duration': 421.044, 'highlights': ['K-means clustering is used in targeted marketing by e-commerce websites like Amazon, Flipkart, and eBay to target the right audience.', 'The process of k-means clustering involves defining the number of clusters (k), choosing centroids for each cluster, computing distances from centroids to data points, and reassigning data points to the closest cluster until centroid values do not change.', 'The ELBO method is used to find the most optimal k value by computing the sum of squared errors for different k values, plotting the k value against the sum of squared errors, and choosing the k value at which the distortion decreases abruptly.']}, {'end': 10576.94, 'start': 10280.743, 'title': 'Image compression with k-means', 'summary': 'Demonstrates the application of k-means clustering to compress an image with 16 million colors down to just 16 colors, resulting in a recognizable but distorted image, showcasing the power of the algorithm.', 'duration': 296.197, 'highlights': ['The image is compressed from 16 million colors to 16 colors using k-means clustering, resulting in a recognizable but distorted image.', 'The mini batch k-means algorithm is used to operate on subsets of the large dataset for quicker and more accurate computation due to the huge number of pixel combinations.', 'Each pixel in the image is considered as a data point, contributing to the creation of matrices, resulting in a very large dataset of 16 million pixels.']}, {'end': 11004.411, 'start': 10578.221, 'title': 'Reinforcement learning basics', 'summary': 'Introduces reinforcement learning, highlighting its role in machine learning, its applications in self-driving cars and deep learning, and its key components including the agent, environment, state, action, reward, policy, value, and action value.', 'duration': 426.19, 'highlights': ['Reinforcement learning is crucial in machine learning, with applications in self-driving cars and deep learning such as AlphaGo.', 'Reinforcement learning involves an agent learning to maximize rewards in an unknown environment through trial and error.', 'Key components of reinforcement learning include the agent, environment, state, action, reward, policy, value, and action value.']}, {'end': 11488.283, 'start': 11004.911, 'title': 'Reinforcement learning concepts', 'summary': 'Discusses key concepts in reinforcement learning such as reward maximization, exploration vs exploitation trade-off, markov decision process, and learning approaches, emphasizing the aim of maximizing rewards and the mathematical mapping of solutions. it also introduces the q-learning algorithm as a significant reinforcement learning algorithm.', 'duration': 483.372, 'highlights': ['The main aim behind reinforcement learning is to maximize the rewards that an agent can get.', 'Exploration involves capturing more information about an environment, while exploitation uses already known information to increase reward.', 'Markov decision process provides a mathematical framework for solving reinforcement learning problems, involving actions, states, rewards, policy, and value.', 'Different reinforcement learning approaches, including policy-based, value-based, and action-based, share the end goal of guiding the agent to acquire the most rewards.', 'Q-learning algorithm is introduced as a significant algorithm in reinforcement learning, to be further discussed and implemented using Python.']}], 'duration': 1629.064, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk9859219.jpg', 'highlights': ['K-means clustering used in targeted marketing by e-commerce websites like Amazon, Flipkart, and eBay', 'Reinforcement learning crucial in machine learning, with applications in self-driving cars and deep learning such as AlphaGo', 'Image compressed from 16 million colors to 16 colors using k-means clustering', 'ELBO method used to find the most optimal k value by computing the sum of squared errors for different k values']}, {'end': 12669.68, 'segs': [{'end': 11520.243, 'src': 'embed', 'start': 11488.823, 'weight': 3, 'content': [{'end': 11491.465, 'text': 'So this is how our demonstration looks for now.', 'start': 11488.823, 'duration': 2.642}, {'end': 11498.729, 'text': 'Now the problem statement is to place an agent in any one of the rooms numbered zero, one, two, three and four,', 'start': 11491.965, 'duration': 6.764}, {'end': 11503.111, 'text': 'and the goal is for the agent to reach outside the building, which is room number five.', 'start': 11498.729, 'duration': 4.382}, {'end': 11510.635, 'text': 'So basically this zero, one, two, three, four represents the building and five represents a room which is outside the building.', 'start': 11503.709, 'duration': 6.926}, {'end': 11513.798, 'text': 'Now all these rooms are connected by doors.', 'start': 11511.236, 'duration': 2.562}, {'end': 11520.243, 'text': 'Now these gaps that you see between the rooms are basically doors and each room is numbered from zero to four.', 'start': 11514.278, 'duration': 5.965}], 'summary': 'Demonstration aims to move agent from room 0-4 to room 5.', 'duration': 31.42, 'max_score': 11488.823, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11488823.jpg'}, {'end': 11677.403, 'src': 'embed', 'start': 11651.889, 'weight': 4, 'content': [{'end': 11656.95, 'text': 'Now, of course, room number five will loop back to itself with a reward of 100,', 'start': 11651.889, 'duration': 5.061}, {'end': 11661.751, 'text': 'and all other direct connections to the gold room will carry a reward of 100..', 'start': 11656.95, 'duration': 4.801}, {'end': 11669.453, 'text': 'Now in Q-learning, the goal is to reach the state with the highest reward so that if the agent arrives at the goal, it will remain there forever.', 'start': 11661.751, 'duration': 7.702}, {'end': 11672.099, 'text': 'So I hope all of you are clear with this diagram.', 'start': 11670.057, 'duration': 2.042}, {'end': 11677.403, 'text': 'Now the terminologies in Q learning include two terms, state and action.', 'start': 11672.799, 'duration': 4.604}], 'summary': 'Q-learning aims to reach state with highest reward, 100 reward for room 5 loop and direct connections to gold room.', 'duration': 25.514, 'max_score': 11651.889, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11651889.jpg'}, {'end': 11742.859, 'src': 'embed', 'start': 11715.918, 'weight': 2, 'content': [{'end': 11720.161, 'text': 'If you go to state four, from there you can directly go to your goal room which is five.', 'start': 11715.918, 'duration': 4.243}, {'end': 11722.644, 'text': 'This is how the agent is going to traverse.', 'start': 11720.782, 'duration': 1.862}, {'end': 11729.035, 'text': "Now in order to depict the rewards that you're going to get we're going to create a matrix known as the reward matrix.", 'start': 11723.354, 'duration': 5.681}, {'end': 11733.357, 'text': 'Okay this is represented by R or also known as the R matrix.', 'start': 11729.536, 'duration': 3.821}, {'end': 11737.578, 'text': 'Now the minus one in this table represents null values.', 'start': 11733.937, 'duration': 3.641}, {'end': 11742.859, 'text': "That is basically where there isn't a link between the nodes that is represented as minus one.", 'start': 11737.998, 'duration': 4.861}], 'summary': 'Agent traverses from state 4 to goal room 5, using reward matrix with -1 for null links.', 'duration': 26.941, 'max_score': 11715.918, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11715918.jpg'}, {'end': 11860.665, 'src': 'embed', 'start': 11821.726, 'weight': 6, 'content': [{'end': 11827.049, 'text': 'Here we have Q state comma action R state comma action, which is nothing but the reward matrix.', 'start': 11821.726, 'duration': 5.323}, {'end': 11832.15, 'text': "then we have a parameter known as the gamma parameter, which I'll explain shortly,", 'start': 11827.529, 'duration': 4.621}, {'end': 11836.992, 'text': 'and then we are multiplying this with the maximum of Q next state comma all actions.', 'start': 11832.15, 'duration': 4.842}, {'end': 11841.233, 'text': "Now don't worry if you haven't understood this formula, I'll explain this with a small example.", 'start': 11837.432, 'duration': 3.801}, {'end': 11844.054, 'text': "For now, let's understand what the gamma parameter is.", 'start': 11841.753, 'duration': 2.301}, {'end': 11847.656, 'text': 'So basically the value of gamma will be between zero and one.', 'start': 11844.654, 'duration': 3.002}, {'end': 11854.1, 'text': 'If gamma is closer to zero, it means that the agent will tend to consider only immediate rewards.', 'start': 11848.137, 'duration': 5.963}, {'end': 11860.665, 'text': 'Now if the gamma is closer to one, it means that the agent will consider future rewards with greater weight.', 'start': 11854.601, 'duration': 6.064}], 'summary': 'Reinforcement learning involves reward matrix, gamma parameter, and future rewards.', 'duration': 38.939, 'max_score': 11821.726, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11821726.jpg'}, {'end': 11996.752, 'src': 'embed', 'start': 11967.76, 'weight': 8, 'content': [{'end': 11972.343, 'text': "Now from the current state you'll select some action that will lead you to the next state.", 'start': 11967.76, 'duration': 4.583}, {'end': 11979.197, 'text': "then you'll basically get the maximum Q value for this next state based on all the possible actions that you take.", 'start': 11973.072, 'duration': 6.125}, {'end': 11983.661, 'text': "Then you'll keep computing this Q value until you reach the goal state.", 'start': 11979.618, 'duration': 4.043}, {'end': 11988.565, 'text': "Now that might be a little bit confusing, so let's look at this entire thing with a small example.", 'start': 11984.302, 'duration': 4.263}, {'end': 11992.609, 'text': "Let's say that first you're gonna begin with setting your gamma parameter.", 'start': 11989.186, 'duration': 3.423}, {'end': 11996.752, 'text': "So I've set my gamma parameter to 0.8, which is pretty close to one.", 'start': 11992.949, 'duration': 3.803}], 'summary': 'Iteratively compute q value to reach goal state, using gamma=0.8.', 'duration': 28.992, 'max_score': 11967.76, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11967760.jpg'}, {'end': 12108.386, 'src': 'embed', 'start': 12080.792, 'weight': 5, 'content': [{'end': 12083.754, 'text': 'Again, your reward matrix R one comma five is 100.', 'start': 12080.792, 'duration': 2.962}, {'end': 12087.917, 'text': "So here you're going to put 100 plus your gamma parameter.", 'start': 12083.754, 'duration': 4.163}, {'end': 12090.698, 'text': 'Your gamma parameter is 0.8.', 'start': 12088.037, 'duration': 2.661}, {'end': 12097.101, 'text': "Then you're going to calculate the maximum Q value for the next state based on all possible actions.", 'start': 12090.698, 'duration': 6.403}, {'end': 12099.022, 'text': "So let's look at the next state.", 'start': 12097.661, 'duration': 1.361}, {'end': 12104.004, 'text': 'From room number five you can go to either one, you can go to four, or you can go to five.', 'start': 12099.142, 'duration': 4.862}, {'end': 12108.386, 'text': 'So your actions are five comma one, five comma four, and five comma five.', 'start': 12104.224, 'duration': 4.162}], 'summary': 'Reward matrix r 1,5 is 100, calculate max q value for next state based on all possible actions.', 'duration': 27.594, 'max_score': 12080.792, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12080792.jpg'}, {'end': 12510.102, 'src': 'embed', 'start': 12482.981, 'weight': 1, 'content': [{'end': 12487.465, 'text': 'Next, we need to update our Q matrix depending on the actions that we took, if you remember.', 'start': 12482.981, 'duration': 4.484}, {'end': 12490.707, 'text': "So that's exactly what this update function is for.", 'start': 12488.185, 'duration': 2.522}, {'end': 12495.771, 'text': 'Now guys this entire is for calculating the Q value.', 'start': 12491.227, 'duration': 4.544}, {'end': 12503.217, 'text': 'I hope all of you remember the formula which is Q state comma action R state comma action plus gamma into max value.', 'start': 12496.451, 'duration': 6.766}, {'end': 12507.9, 'text': 'Max value will basically give me the maximum value out of the all possible actions.', 'start': 12503.657, 'duration': 4.243}, {'end': 12510.102, 'text': "I'm basically computing this formula.", 'start': 12508.501, 'duration': 1.601}], 'summary': 'Updating q matrix to calculate q value using formula', 'duration': 27.121, 'max_score': 12482.981, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12482981.jpg'}, {'end': 12554.399, 'src': 'embed', 'start': 12526.385, 'weight': 0, 'content': [{'end': 12531.247, 'text': 'You can set this depending on your own needs and 10, 000 iterations is a pretty huge number.', 'start': 12526.385, 'duration': 4.862}, {'end': 12537.53, 'text': 'So basically my agent is going to go through 10, 000 possible iterations in order to find the best policy.', 'start': 12532.007, 'duration': 5.523}, {'end': 12541.452, 'text': 'Now this is the exact same thing that we did earlier.', 'start': 12538.81, 'duration': 2.642}, {'end': 12545.794, 'text': "We're setting the current state and then we're choosing the available action from the current state.", 'start': 12541.672, 'duration': 4.122}, {'end': 12548.515, 'text': "Then from there we'll choose a action at random.", 'start': 12546.214, 'duration': 2.301}, {'end': 12554.399, 'text': "Here we'll calculate the Q value and we'll update the Q value in the matrix.", 'start': 12549.016, 'duration': 5.383}], 'summary': 'Agent will go through 10,000 iterations to find best policy.', 'duration': 28.014, 'max_score': 12526.385, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12526385.jpg'}], 'start': 11488.823, 'title': 'Reinforcement learning in a building', 'summary': 'Explores placing an agent in a building with rooms numbered 0 to 4, using q-learning to navigate to room 5 with direct and indirect links, rewards, and emphasizes the significance of the gamma parameter in balancing exploitation and exploration in the q-learning algorithm, and demonstrates the implementation of reinforcement learning using the numpy library through 10,000 iterations.', 'chapters': [{'end': 11753.542, 'start': 11488.823, 'title': 'Reinforcement learning in a building', 'summary': 'Discusses placing an agent in a building with rooms numbered 0 to 4 and the goal for the agent to reach outside the building, room number 5. the agent will use q-learning to traverse the rooms, with direct links from rooms 1 and 4 to 5, and indirect links between other rooms, and rewards of 100 for direct links to the goal, and 0 for other links.', 'duration': 264.719, 'highlights': ["The agent's goal is to reach room number 5, with direct links from rooms 1 and 4 to 5, and indirect links between other rooms, illustrating the traversal challenge the agent faces.", 'In Q-learning, the goal is to reach the state with the highest reward, with direct links to the goal room carrying a reward of 100 and other doors having a reward of 0, emphasizing the reinforcement learning approach for the agent to reach the goal state.', "The reward matrix, represented by R, depicts the rewards the agent will receive for traversing between different states, with null values represented as -1 and direct links between rooms carrying specific rewards, providing a visual representation of the rewards associated with the agent's movements."]}, {'end': 12283.102, 'start': 11753.963, 'title': 'Q-learning algorithm', 'summary': 'Explains the reward matrix, q matrix, and gamma parameter in q-learning algorithm, emphasizing the significance of gamma parameter in balancing exploitation and exploration, and describes the step-by-step process of implementing q-learning algorithm with an example.', 'duration': 529.139, 'highlights': ['Explanation of Q matrix and reward matrix', 'Significance of gamma parameter in Q-learning', 'Step-by-step process of Q-learning algorithm with an example']}, {'end': 12669.68, 'start': 12284.051, 'title': 'Reinforcement learning with numpy', 'summary': 'Demonstrates the implementation of reinforcement learning using the numpy library, creating and updating r and q matrices, setting the gamma parameter, training the agent through 10,000 iterations, and testing the best policy to reach the goal state, room number five.', 'duration': 385.629, 'highlights': ['The agent performs 10,000 iterations during the training phase to find the best policy.', 'The Q matrix is updated based on the formula Q(state, action) = R(state, action) + gamma * max value.', 'The process involves creating R and Q matrices, setting the gamma parameter, and selecting the best policy to reach the goal state.']}], 'duration': 1180.857, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk11488823.jpg', 'highlights': ['The agent performs 10,000 iterations during the training phase to find the best policy.', 'The Q matrix is updated based on the formula Q(state, action) = R(state, action) + gamma * max value.', "The reward matrix, represented by R, depicts the rewards the agent will receive for traversing between different states, with null values represented as -1 and direct links between rooms carrying specific rewards, providing a visual representation of the rewards associated with the agent's movements.", "The agent's goal is to reach room number 5, with direct links from rooms 1 and 4 to 5, and indirect links between other rooms, illustrating the traversal challenge the agent faces.", 'In Q-learning, the goal is to reach the state with the highest reward, with direct links to the goal room carrying a reward of 100 and other doors having a reward of 0, emphasizing the reinforcement learning approach for the agent to reach the goal state.', 'The process involves creating R and Q matrices, setting the gamma parameter, and selecting the best policy to reach the goal state.', 'Explanation of Q matrix and reward matrix', 'Significance of gamma parameter in Q-learning', 'Step-by-step process of Q-learning algorithm with an example']}, {'end': 13804.319, 'segs': [{'end': 12823.216, 'src': 'embed', 'start': 12729.668, 'weight': 1, 'content': [{'end': 12737.311, 'text': 'Now artificial intelligence is basically the science of getting machines to mimic the behavior of human beings.', 'start': 12729.668, 'duration': 7.643}, {'end': 12739.751, 'text': 'But when it comes to machine learning,', 'start': 12737.731, 'duration': 2.02}, {'end': 12748.314, 'text': 'machine learning is a subset of artificial intelligence that focuses on getting machines to make decisions by feeding them data.', 'start': 12739.751, 'duration': 8.563}, {'end': 12750.735, 'text': "That's exactly what machine learning is.", 'start': 12748.954, 'duration': 1.781}, {'end': 12753.596, 'text': 'It is a subset of artificial intelligence.', 'start': 12751.235, 'duration': 2.361}, {'end': 12761.491, 'text': 'Deep learning, on the other hand, is a subset of machine learning that uses the concept of neural networks to solve complex problems.', 'start': 12754.236, 'duration': 7.255}, {'end': 12767.911, 'text': 'So to sum it up, artificial intelligence, machine learning, and deep learning are interconnected fields.', 'start': 12762.21, 'duration': 5.701}, {'end': 12777.393, 'text': 'Machine learning and deep learning aids artificial intelligence by providing a set of algorithms and neural networks to solve data-driven problems.', 'start': 12768.611, 'duration': 8.782}, {'end': 12781.113, 'text': "That's how AI, machine learning, and deep learning are related.", 'start': 12778.033, 'duration': 3.08}, {'end': 12786.714, 'text': 'I hope all of you have cleared your misconceptions and doubts about AI, ML, and deep learning.', 'start': 12781.793, 'duration': 4.921}, {'end': 12790.995, 'text': "Now let's look at our next topic, which is limitations of machine learning.", 'start': 12787.374, 'duration': 3.621}, {'end': 12797.992, 'text': 'Now the first limitation is machine learning is not capable enough to handle high dimensional data.', 'start': 12791.846, 'duration': 6.146}, {'end': 12801.075, 'text': 'This is where the input and the output is very large.', 'start': 12798.533, 'duration': 2.542}, {'end': 12807.882, 'text': 'So handling and processing such type of data becomes very complex and it takes up a lot of resources.', 'start': 12801.515, 'duration': 6.367}, {'end': 12811.865, 'text': 'This is also sometimes known as the curse of dimensionality.', 'start': 12808.522, 'duration': 3.343}, {'end': 12817.351, 'text': 'So to understand this in simpler terms, look at the image shown on this slide.', 'start': 12812.426, 'duration': 4.925}, {'end': 12823.216, 'text': "Consider a line of 100 yards and let's say that you dropped a coin somewhere on the line.", 'start': 12817.951, 'duration': 5.265}], 'summary': 'Ai, ml, and deep learning are interconnected fields. ml is limited by handling high dimensional data.', 'duration': 93.548, 'max_score': 12729.668, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12729668.jpg'}, {'end': 13024.567, 'src': 'embed', 'start': 12998.095, 'weight': 0, 'content': [{'end': 13004.558, 'text': 'Now deep learning is one of the only methods by which we can overcome the challenges of feature extraction.', 'start': 12998.095, 'duration': 6.463}, {'end': 13011.801, 'text': 'This is because deep learning models are capable of learning to focus on the right features by themselves,', 'start': 13005.058, 'duration': 6.743}, {'end': 13014.302, 'text': 'which requires very little guidance from the programmer.', 'start': 13011.801, 'duration': 2.501}, {'end': 13017.757, 'text': 'Basically, deep learning mimics the way our brain functions.', 'start': 13014.873, 'duration': 2.884}, {'end': 13019.72, 'text': 'That is, it learns from experience.', 'start': 13017.877, 'duration': 1.843}, {'end': 13024.567, 'text': 'So in deep learning, what happens is feature extraction happens automatically.', 'start': 13020.421, 'duration': 4.146}], 'summary': 'Deep learning automates feature extraction, mimics brain functions, and requires little guidance from programmers.', 'duration': 26.472, 'max_score': 12998.095, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12998095.jpg'}], 'start': 12670.54, 'title': 'Ai, ml, and deep learning', 'summary': 'Explores the relationship between artificial intelligence, machine learning, and deep learning, emphasizing the limitations of machine learning in handling high dimensional data, and introducing deep learning as a solution that automates feature extraction and mimics the human brain through artificial neural networks.', 'chapters': [{'end': 12877.661, 'start': 12670.54, 'title': 'Ai, ml, and deep learning', 'summary': 'Explains the relationship between artificial intelligence, machine learning, and deep learning, and discusses the limitations of machine learning, highlighting the curse of dimensionality and its impact on handling high dimensional data.', 'duration': 207.121, 'highlights': ['Artificial intelligence is the science of getting machines to mimic human behavior, while machine learning is a subset of artificial intelligence that focuses on decision making based on data.', 'Deep learning uses neural networks to solve complex problems and is a subset of machine learning.', 'The curse of dimensionality is a limitation of machine learning, making it challenging to handle high dimensional data due to increased complexity and resource consumption.']}, {'end': 13804.319, 'start': 12878.201, 'title': 'Limitations of machine learning and introduction to deep learning', 'summary': 'Highlights the limitations of machine learning in handling high dimensional data, the challenges of feature extraction and the introduction of deep learning as a solution, illustrating how deep learning automates feature extraction and mimics the functioning of the human brain through artificial neural networks.', 'duration': 926.118, 'highlights': ['Deep learning automates feature extraction and mimics the functioning of the human brain through artificial neural networks', 'Limitations of machine learning in handling high dimensional data', 'Challenges of feature extraction in machine learning']}], 'duration': 1133.779, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk12670540.jpg', 'highlights': ['Deep learning automates feature extraction and mimics the functioning of the human brain through artificial neural networks', 'Artificial intelligence is the science of getting machines to mimic human behavior, while machine learning is a subset of artificial intelligence that focuses on decision making based on data', 'Deep learning uses neural networks to solve complex problems and is a subset of machine learning', 'The curse of dimensionality is a limitation of machine learning, making it challenging to handle high dimensional data due to increased complexity and resource consumption', 'Limitations of machine learning in handling high dimensional data', 'Challenges of feature extraction in machine learning']}, {'end': 14701.58, 'segs': [{'end': 13882.007, 'src': 'embed', 'start': 13855.206, 'weight': 1, 'content': [{'end': 13858.73, 'text': "So now we'll discuss something known as multilayer perceptron.", 'start': 13855.206, 'duration': 3.524}, {'end': 13866.078, 'text': 'A multilayer perceptron has the same structure of a single layer perceptron, but with one or more hidden layers.', 'start': 13859.33, 'duration': 6.748}, {'end': 13869.823, 'text': "Okay, and that's why it's considered as a deep neural network.", 'start': 13866.559, 'duration': 3.264}, {'end': 13874.601, 'text': 'So in a single layer perceptron, we had only an input layer, output layer.', 'start': 13870.558, 'duration': 4.043}, {'end': 13876.402, 'text': "We didn't have any hidden layer.", 'start': 13875.001, 'duration': 1.401}, {'end': 13882.007, 'text': 'Now when it comes to multi-layer perceptrons, there are hidden layers in between, and then there is the output layer.', 'start': 13876.823, 'duration': 5.184}], 'summary': 'Multilayer perceptrons have hidden layers, making them deep neural networks.', 'duration': 26.801, 'max_score': 13855.206, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk13855206.jpg'}, {'end': 14063.358, 'src': 'heatmap', 'start': 13876.823, 'weight': 0.702, 'content': [{'end': 13882.007, 'text': 'Now when it comes to multi-layer perceptrons, there are hidden layers in between, and then there is the output layer.', 'start': 13876.823, 'duration': 5.184}, {'end': 13884.008, 'text': 'It works in a similar manner.', 'start': 13882.627, 'duration': 1.381}, {'end': 13891.053, 'text': "like I said, first you'll have the inputs X1, X2, X3, and so on, and each of these inputs will be assigned some weight, right?", 'start': 13884.008, 'duration': 7.045}, {'end': 13893.275, 'text': 'W1, W2, W3, and so on.', 'start': 13891.434, 'duration': 1.841}, {'end': 13898.199, 'text': "Then you'll calculate the weighted summation of each of these inputs and their weights.", 'start': 13893.776, 'duration': 4.423}, {'end': 13903.289, 'text': "After that, you send them to the transformation or the activation function and you'll finally get the output.", 'start': 13898.906, 'duration': 4.383}, {'end': 13908.153, 'text': "Now the only thing is that you'll have multiple hidden layers in between.", 'start': 13904.05, 'duration': 4.103}, {'end': 13910.374, 'text': 'One or more than one hidden layers.', 'start': 13908.813, 'duration': 1.561}, {'end': 13913.757, 'text': 'So guys this is how a multilayer perceptron works.', 'start': 13911.075, 'duration': 2.682}, {'end': 13917.62, 'text': 'It works on the concept of feed forward neural networks.', 'start': 13914.057, 'duration': 3.563}, {'end': 13923.804, 'text': 'Feed forward means every node at each level or each layer is connected to every other node.', 'start': 13918.2, 'duration': 5.604}, {'end': 13926.026, 'text': "So that's what feed forward networks are.", 'start': 13924.345, 'duration': 1.681}, {'end': 13931.152, 'text': 'Now when it comes to assigning weights, what we do is we randomly assign weights.', 'start': 13926.75, 'duration': 4.402}, {'end': 13933.734, 'text': 'Initially we have input X1, X2, X3.', 'start': 13931.232, 'duration': 2.502}, {'end': 13937.476, 'text': 'We randomly assign some weight W1, W2, W3 and so on.', 'start': 13933.754, 'duration': 3.722}, {'end': 13945.64, 'text': "Now, it's always necessary that, whatever weights we assign to our input, those weights are actually correct,", 'start': 13938.196, 'duration': 7.444}, {'end': 13949.962, 'text': 'meaning that those weights are actually significant in predicting your output.', 'start': 13945.64, 'duration': 4.322}, {'end': 13956.25, 'text': 'So how a multilayer perceptron works is a set of inputs are passed to the first hidden layer.', 'start': 13950.925, 'duration': 5.325}, {'end': 13965.339, 'text': "Now the activations from that layer are passed to the next layer and from that layer it's passed to the next hidden layer until you reach the output layer.", 'start': 13956.27, 'duration': 9.069}, {'end': 13969.504, 'text': "From the output layer, you'll form the two classes, class one and class two.", 'start': 13966.14, 'duration': 3.364}, {'end': 13973.628, 'text': "Basically, you'll classify your input into one of the two classes.", 'start': 13969.824, 'duration': 3.804}, {'end': 13976.229, 'text': "So that's how a multilayer perceptron works.", 'start': 13974.248, 'duration': 1.981}, {'end': 13981.391, 'text': 'A very important concept in multiple layer perceptron is back propagation.', 'start': 13976.809, 'duration': 4.582}, {'end': 13989.194, 'text': 'Now what is back propagation? Back propagation algorithm is a supervised learning method for multilayer perceptrons.', 'start': 13982.011, 'duration': 7.183}, {'end': 13992.135, 'text': 'Okay, now, why do we need back propagation?', 'start': 13989.734, 'duration': 2.401}, {'end': 14000.958, 'text': 'So, guys, when we are designing a neural network, in the beginning we initialize weights with some random values or any variable for that fact.', 'start': 13992.775, 'duration': 8.183}, {'end': 14005.973, 'text': 'Now, obviously, we need to make sure that these weights actually are correct,', 'start': 14001.511, 'duration': 4.462}, {'end': 14010.276, 'text': 'meaning that these weights show the significance of each predictor variable.', 'start': 14005.973, 'duration': 4.303}, {'end': 14015.118, 'text': 'These weights have to fit our model in such a way that our output is very precise.', 'start': 14010.776, 'duration': 4.342}, {'end': 14023.543, 'text': "So let's say that we randomly selected some weights in the beginning, but our model output is much more different than our actual output,", 'start': 14015.638, 'duration': 7.905}, {'end': 14025.644, 'text': 'meaning that our error value is very huge.', 'start': 14023.543, 'duration': 2.101}, {'end': 14028.045, 'text': 'So how will you reduce this error??', 'start': 14026.264, 'duration': 1.781}, {'end': 14037.364, 'text': 'Basically, what you need to do is we need to somehow explain to the model that we need to change the weights in such a way that the error becomes minimum.', 'start': 14028.65, 'duration': 8.714}, {'end': 14042.252, 'text': 'So the main thing is the weights and your error is very highly related.', 'start': 14037.965, 'duration': 4.287}, {'end': 14048.032, 'text': 'the weightage that you give to each input will show how much error is there in your output,', 'start': 14042.93, 'duration': 5.102}, {'end': 14054.935, 'text': 'because the most significant variables will have the highest weightage, and if the weightage is not correct, then your output is also not correct.', 'start': 14048.032, 'duration': 6.903}, {'end': 14063.358, 'text': 'Now back propagation is a way to update your weights in such a way that your outcome is precise and your error is reduced.', 'start': 14055.375, 'duration': 7.983}], 'summary': 'Multilayer perceptrons use hidden layers and back propagation for precise output and error reduction.', 'duration': 186.535, 'max_score': 13876.823, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk13876823.jpg'}, {'end': 14094.389, 'src': 'embed', 'start': 14064.071, 'weight': 0, 'content': [{'end': 14068.519, 'text': 'So in short, back propagation is used to train a multilayer perceptron.', 'start': 14064.071, 'duration': 4.448}, {'end': 14076.873, 'text': "It's basically used to update your weights in such a way that your output is more precise and that your error is reduced.", 'start': 14069.08, 'duration': 7.793}, {'end': 14080.803, 'text': 'So training a neural network is all about back propagation.', 'start': 14077.662, 'duration': 3.141}, {'end': 14087.906, 'text': 'So the most common deep learning algorithm for supervised training of the multilayer perceptron is known as back propagation.', 'start': 14081.204, 'duration': 6.702}, {'end': 14094.389, 'text': 'So, after calculating the weighted sum of inputs and passing them through the activation function,', 'start': 14088.467, 'duration': 5.922}], 'summary': 'Back propagation trains multilayer perceptron for more precise output and error reduction.', 'duration': 30.318, 'max_score': 14064.071, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk14064071.jpg'}, {'end': 14585.826, 'src': 'heatmap', 'start': 14405.707, 'weight': 0.735, 'content': [{'end': 14410.97, 'text': "Now in order to tell you mathematically what we're doing is we're using a method known as gradient descent.", 'start': 14405.707, 'duration': 5.263}, {'end': 14418.415, 'text': 'Okay, this method is used to adjust all the weights in the network with an aim of reducing the error at the output layer.', 'start': 14411.37, 'duration': 7.045}, {'end': 14426.299, 'text': 'So how gradient descent optimizer works is the first step is you will calculate the error by considering the below equation.', 'start': 14419.035, 'duration': 7.264}, {'end': 14430.742, 'text': "Here you're subtracting the summation of your actual output from your network output.", 'start': 14426.84, 'duration': 3.902}, {'end': 14438.637, 'text': 'Step two is based on the error you get, you will calculate the rate of change of error with respect to the change in the weight.', 'start': 14431.434, 'duration': 7.203}, {'end': 14442.939, 'text': 'The learning rate is something that you set in the beginning itself.', 'start': 14439.318, 'duration': 3.621}, {'end': 14447.981, 'text': 'Step three is based on this change in weight, you will calculate the new weight.', 'start': 14443.679, 'duration': 4.302}, {'end': 14452.703, 'text': 'Alright, your updated weight will be your weight plus the rate of change of weight.', 'start': 14448.462, 'duration': 4.241}, {'end': 14456.845, 'text': 'So guys, that was all about back propagation and weight update.', 'start': 14453.344, 'duration': 3.501}, {'end': 14460.827, 'text': "Now let's look at the limitations of feed forward network.", 'start': 14457.465, 'duration': 3.362}, {'end': 14466.502, 'text': 'So far, we were discussing the multiple layer perceptron, which uses the feedforward network.', 'start': 14461.46, 'duration': 5.042}, {'end': 14469.884, 'text': "Now, let's discuss the limitations of these feedforward networks.", 'start': 14466.942, 'duration': 2.942}, {'end': 14473.926, 'text': "Now, let's consider an example of image classification.", 'start': 14470.724, 'duration': 3.202}, {'end': 14479.028, 'text': "Okay, let's say you've trained the neural network to classify images of various animals.", 'start': 14474.226, 'duration': 4.802}, {'end': 14480.969, 'text': "Now, let's consider an example.", 'start': 14479.488, 'duration': 1.481}, {'end': 14490.713, 'text': 'Here, the first output is an elephant, right? We have an elephant, and this output will have nothing to do with the previous output, which is a dog.', 'start': 14481.549, 'duration': 9.164}, {'end': 14497.33, 'text': 'This means that the output at time t is independent of the output at time t minus one.', 'start': 14491.443, 'duration': 5.887}, {'end': 14503.318, 'text': 'Now consider the scenario where you will require the use of previously obtained output.', 'start': 14497.931, 'duration': 5.387}, {'end': 14506.562, 'text': 'Okay, the concept is very similar to reading a book.', 'start': 14504.119, 'duration': 2.443}, {'end': 14510.847, 'text': 'As you turn every page, you need an understanding of the previous pages.', 'start': 14507.202, 'duration': 3.645}, {'end': 14515.127, 'text': 'If you want to make sense of the information, then you need to know what you learned before.', 'start': 14511.545, 'duration': 3.582}, {'end': 14517.328, 'text': "It's exactly what you're doing right now.", 'start': 14515.667, 'duration': 1.661}, {'end': 14521.85, 'text': 'In order to understand deep learning, you had to understand machine learning.', 'start': 14517.708, 'duration': 4.142}, {'end': 14532.916, 'text': 'So basically with a feed forward network, the new output at time t plus one has nothing to do with the output at time t or t minus one or t minus two.', 'start': 14522.491, 'duration': 10.425}, {'end': 14537.618, 'text': 'So feed forward networks cannot be used while predicting a word in a sentence,', 'start': 14533.456, 'duration': 4.162}, {'end': 14541.1, 'text': 'as it will have absolutely no relationship with the previous set of words.', 'start': 14537.618, 'duration': 3.482}, {'end': 14549.347, 'text': 'So a feed forward network cannot be used in use cases wherein you have to predict the outcome based on your previous outcome.', 'start': 14541.801, 'duration': 7.546}, {'end': 14555.932, 'text': 'So in a lot of use cases, your previous output will also determine your next output.', 'start': 14550.068, 'duration': 5.864}, {'end': 14559.935, 'text': 'So for such cases, you cannot make use of feed forward network.', 'start': 14556.453, 'duration': 3.482}, {'end': 14567.061, 'text': 'Now what modification can you make so that your network can learn from your previous mistakes? For this, we have a solution.', 'start': 14560.376, 'duration': 6.685}, {'end': 14570.492, 'text': 'So a solution to this is recurrent neural networks.', 'start': 14567.91, 'duration': 2.582}, {'end': 14577.298, 'text': "So basically let's say you have an input at time t minus one and you'll get some output when you feed it to the network.", 'start': 14570.973, 'duration': 6.325}, {'end': 14585.826, 'text': 'Now some information from this input at t minus one is fed to the next input which is input at time t.', 'start': 14577.899, 'duration': 7.927}], 'summary': 'Using gradient descent to update weights in neural network, limitations of feedforward networks, and solution with recurrent neural networks.', 'duration': 180.119, 'max_score': 14405.707, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk14405707.jpg'}, {'end': 14649.525, 'src': 'embed', 'start': 14625.832, 'weight': 4, 'content': [{'end': 14633.935, 'text': 'In time series and in stock markets, the main networks that are used are recurrent neural networks because each of your inputs are correlated.', 'start': 14625.832, 'duration': 8.103}, {'end': 14639.357, 'text': "Now to better understand recurrent neural networks, let's consider a small example.", 'start': 14634.634, 'duration': 4.723}, {'end': 14645.522, 'text': "Let's say that you go to the gym regularly and the trainer has given you a schedule for your workout.", 'start': 14639.898, 'duration': 5.624}, {'end': 14649.525, 'text': 'So basically the exercises are repeated after every third day.', 'start': 14646.222, 'duration': 3.303}], 'summary': 'Recurrent neural networks are used in time series and stock markets due to correlated inputs. illustrated with a gym example.', 'duration': 23.693, 'max_score': 14625.832, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk14625832.jpg'}], 'start': 13804.959, 'title': 'Neural networks', 'summary': 'Discusses perceptron, multilayer perceptron, back propagation, weight update, feed forward and recurrent neural networks, and their applications, providing insights into their working principles and limitations, with specific examples and supervised learning methods.', 'chapters': [{'end': 14000.958, 'start': 13804.959, 'title': 'Perceptron and multilayer perceptron', 'summary': 'Explains the working of a perceptron, its limitations, and introduces the concept of multilayer perceptron, including the structure, feed forward neural networks, weight assignment, and back propagation as a supervised learning method.', 'duration': 195.999, 'highlights': ['The most weightage is associated with the input in predicting the output, which is crucial for understanding the working of a perceptron.', 'A multilayer perceptron has one or more hidden layers and is considered a deep neural network, addressing the limitations of a single layer perceptron in solving complex problems involving a lot of parameters.', 'The structure of a multilayer perceptron includes hidden layers between input and output layers, and it works on the concept of feed forward neural networks where every node at each level is connected to every other node.', "The process of weight assignment in a multilayer perceptron involves randomly assigning weights to the inputs, and ensuring that these weights are significant in predicting the output is crucial for the network's functioning.", 'Back propagation is a supervised learning method for multilayer perceptrons, crucial for adjusting the weights during the training process of a neural network.']}, {'end': 14701.58, 'start': 14001.511, 'title': 'Back propagation and weight update', 'summary': 'Explains how back propagation is used to update weights in a multilayer perceptron to minimize error, using a specific example and discussing limitations of feedforward networks, leading to a discussion on recurrent neural networks and their applications in various domains.', 'duration': 700.069, 'highlights': ['Back propagation is used to train a multilayer perceptron and update weights to minimize error, which is crucial in deep learning for supervised training.', 'Explains the relationship between weights and error, demonstrating how increasing or decreasing weights affects the error, and the process of updating weights to minimize error.', 'Describes the use of gradient descent to adjust weights in the network with the goal of reducing error at the output layer, outlining the steps involved in this optimization method.', 'Discusses the limitations of feedforward networks in scenarios where predictions depend on previous outcomes, leading to the introduction of recurrent neural networks as a solution for learning from previous data in sequence-based tasks.', 'Highlights the importance of recurrent neural networks in various domains such as time series analysis and stock markets due to their ability to recognize patterns in correlated inputs.']}], 'duration': 896.621, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk13804959.jpg', 'highlights': ['Back propagation is a supervised learning method for multilayer perceptrons, crucial for adjusting the weights during the training process of a neural network.', 'A multilayer perceptron has one or more hidden layers and is considered a deep neural network, addressing the limitations of a single layer perceptron in solving complex problems involving a lot of parameters.', 'The structure of a multilayer perceptron includes hidden layers between input and output layers, and it works on the concept of feed forward neural networks where every node at each level is connected to every other node.', 'Back propagation is used to train a multilayer perceptron and update weights to minimize error, which is crucial in deep learning for supervised training.', 'Highlights the importance of recurrent neural networks in various domains such as time series analysis and stock markets due to their ability to recognize patterns in correlated inputs.']}, {'end': 17558.523, 'segs': [{'end': 14726.129, 'src': 'embed', 'start': 14702.161, 'weight': 2, 'content': [{'end': 14709.863, 'text': 'So if a model is trained based on the data it can obtain from the previous exercise, the output from the model will be extremely accurate.', 'start': 14702.161, 'duration': 7.702}, {'end': 14716.906, 'text': 'In such cases, you need to know the output at t minus one in order to predict the output at t.', 'start': 14710.504, 'duration': 6.402}, {'end': 14719.927, 'text': 'In such cases, recurrent neural networks are very essential.', 'start': 14716.906, 'duration': 3.021}, {'end': 14726.129, 'text': "So basically on feeding some inputs to the neural networks, you'll go through a few functions and you'll get the output.", 'start': 14720.587, 'duration': 5.542}], 'summary': 'Recurrent neural networks use previous data for accurate predictions.', 'duration': 23.968, 'max_score': 14702.161, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk14702161.jpg'}, {'end': 15093.173, 'src': 'embed', 'start': 15057.448, 'weight': 4, 'content': [{'end': 15059.609, 'text': 'It does not make sense to shuffle this data.', 'start': 15057.448, 'duration': 2.161}, {'end': 15063.492, 'text': "Now the next step we're going to do is we're going to scale the data.", 'start': 15060.29, 'duration': 3.202}, {'end': 15068.986, 'text': 'Now, scaling data and data normalization is one of the most important steps.', 'start': 15064.105, 'duration': 4.881}, {'end': 15070.527, 'text': 'Alright, you cannot miss this step.', 'start': 15069.226, 'duration': 1.301}, {'end': 15074.568, 'text': 'I already mentioned earlier what normalization and scaling is.', 'start': 15071.007, 'duration': 3.561}, {'end': 15078.689, 'text': 'Now, most neural networks benefit from scaling inputs.', 'start': 15075.208, 'duration': 3.481}, {'end': 15085.071, 'text': "This is because most common activation functions of the network's neurons, such as tan, H, and sigmoid.", 'start': 15078.929, 'duration': 6.142}, {'end': 15093.173, 'text': 'Tan, H, and sigmoid are basically activation functions, and these are defined in the range of minus one to one, or zero and one.', 'start': 15085.731, 'duration': 7.442}], 'summary': 'Scaling and normalization are crucial for neural networks, benefiting from inputs in the range of -1 to 1 or 0 to 1.', 'duration': 35.725, 'max_score': 15057.448, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk15057448.jpg'}, {'end': 16112.25, 'src': 'embed', 'start': 16082.606, 'weight': 0, 'content': [{'end': 16087.369, 'text': "Now the final thing that's getting calculated is our error, MSC, or mean squared error.", 'start': 16082.606, 'duration': 4.763}, {'end': 16090.652, 'text': "So guys, don't worry about this warning, right, it's just a warning.", 'start': 16087.95, 'duration': 2.702}, {'end': 16098.198, 'text': 'So our mean squared error comes down to 0.0029, which is pretty low because the target is scaled.', 'start': 16091.192, 'duration': 7.006}, {'end': 16101.16, 'text': 'and this means that our accuracy is pretty good.', 'start': 16098.878, 'duration': 2.282}, {'end': 16107.225, 'text': 'So, guys, like I mentioned, if you want to improve the accuracy of the model, you can use different schemes,', 'start': 16101.761, 'duration': 5.464}, {'end': 16112.25, 'text': 'you can use different initialization functions or you can try out different transformation functions.', 'start': 16107.225, 'duration': 5.025}], 'summary': 'Mean squared error is 0.0029, indicating good accuracy in the model.', 'duration': 29.644, 'max_score': 16082.606, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk16082606.jpg'}, {'end': 16422.993, 'src': 'embed', 'start': 16397.174, 'weight': 3, 'content': [{'end': 16403.961, 'text': 'NLP uses concepts of computer science and artificial intelligence to study the data and derive useful information from it.', 'start': 16397.174, 'duration': 6.787}, {'end': 16409.622, 'text': "Now before we move any further, let's look at a few applications of NLP and text mining.", 'start': 16404.597, 'duration': 5.025}, {'end': 16413.285, 'text': 'Now we all spend a lot of time surfing the web.', 'start': 16410.241, 'duration': 3.044}, {'end': 16420.03, 'text': 'Have you ever noticed that if you start typing a word on Google, you immediately get suggestions like these.', 'start': 16413.845, 'duration': 6.185}, {'end': 16422.993, 'text': 'This feature is also known as autocomplete.', 'start': 16420.651, 'duration': 2.342}], 'summary': 'Nlp uses ai to derive info from data. nlp has applications in text mining and autocomplete features on google.', 'duration': 25.819, 'max_score': 16397.174, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk16397174.jpg'}, {'end': 17078.652, 'src': 'embed', 'start': 17053.26, 'weight': 1, 'content': [{'end': 17060.524, 'text': "After this, we'll be using the naive bias classifier and we'll define the object for the naive bias classifier, which is basically classifier,", 'start': 17053.26, 'duration': 7.264}, {'end': 17062.665, 'text': "and we'll train this using our training data set.", 'start': 17060.524, 'duration': 2.141}, {'end': 17065.547, 'text': "We'll also look at the accuracy of our model.", 'start': 17063.265, 'duration': 2.282}, {'end': 17069.769, 'text': 'The accuracy of our classifier is around 73% which is a really good number.', 'start': 17066.187, 'duration': 3.582}, {'end': 17078.652, 'text': 'Now this classifier object will actually contain the most informative words that are obtained during analysis.', 'start': 17072.631, 'duration': 6.021}], 'summary': 'Using the naive bias classifier, trained with a 73% accuracy, to identify informative words.', 'duration': 25.392, 'max_score': 17053.26, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk17053260.jpg'}], 'start': 14702.161, 'title': 'Deep learning and nlp', 'summary': 'Covers the use of neural networks, deep learning for stock price prediction, tensorflow training achieving a mean squared error of 0.0029, and the importance of text mining and nlp, featuring a 73% accuracy in sentimental analysis using naive bias classifier.', 'chapters': [{'end': 14932.837, 'start': 14702.161, 'title': 'Understanding neural networks', 'summary': 'Explains the use of recurrent neural networks to predict outputs based on past information, and the necessity of convolutional neural networks to address overfitting by connecting only significant regions of the layer before it.', 'duration': 230.676, 'highlights': ['Recurrent neural networks are essential for predicting outputs based on past information.', 'Importance of convolutional neural networks in addressing overfitting by connecting only significant regions of the layer before it.', 'Explanation of how computers process images and the need for convolutional neural networks in handling image data.']}, {'end': 15693.792, 'start': 14933.238, 'title': 'Deep learning for stock price prediction', 'summary': 'Demonstrates using deep learning to predict stock prices with a practical demo using a data set containing 42,000 minutes of data on 500 stocks and the s&p 500 index from april to august 2017, preparing the data by dropping unnecessary variables, splicing the data for training and testing, and emphasizing the importance of scaling and normalization of data.', 'duration': 760.554, 'highlights': ['The data set contains around 42,000 minutes of data ranging from April to August 2017 on 500 stocks as well as the total S&P 500 index price.', 'Preparing the training and the testing data with the training data containing 80% of the total data set and not shuffling the data set for sequential slicing.', 'Emphasizing the importance of scaling and normalization of data for neural networks, using min-max scaler and cautioning against scaling the whole data set before training and test splits.', 'Introduction and importance of TensorFlow for deep learning and neural network computation framework.', "Explanation of placeholders for storing input and target data in the model architecture, with 'x' containing the network's input and 'y' containing the network's output."]}, {'end': 16139.552, 'start': 15694.292, 'title': 'Deep learning in tensorflow', 'summary': 'Explains the process of mini-batch training in deep learning using tensorflow, involving random sampling of data, epoch training, and the evaluation of model predictions, with a mean squared error of 0.0029 achieved.', 'duration': 445.26, 'highlights': ['The mini-batch training method involves randomly sampling data from a large dataset and sequentially feeding it into the network, with each batch comprising 256 samples.', 'One full sweep over all batches is known as an epoch, and the process is repeated 10 times to train the model.', "During training, the network's predictions on the test set are evaluated every fifth batch, and the mean squared error achieved is 0.0029, indicating a good level of accuracy.", 'Different methods for improving model accuracy are mentioned, including changing the design of layers, using different initialization and activation functions, implementing dropout layers, and early stopping techniques.', 'The chapter concludes with a recap of the deep learning demo and an invitation for any further questions or clarifications in the comments section.']}, {'end': 16819.125, 'start': 16140.253, 'title': 'Importance of text mining and nlp', 'summary': 'Discusses the need for text mining and natural language processing due to the vast amount of unstructured data generated daily, highlighting examples of data volume, applications, and key concepts such as tokenization, stemming, and lemmatization.', 'duration': 678.872, 'highlights': ['Around 2.5 quintillion bytes of data is created every day, with unstructured data accounting for 79%', '1.7 million pictures are posted on Instagram every minute, along with around 347,000 tweets on Twitter', 'Applications of NLP and text mining include autocomplete, spam detection, predictive typing, sentiment analysis, chatbots, speech recognition, machine translation, and advertisement matching', 'Tokenization involves breaking down data into smaller chunks for analysis', 'Stemming and lemmatization are methods of normalizing words, with lemmatization considering morphological analysis for accurate results']}, {'end': 17558.523, 'start': 16819.705, 'title': 'Nlp: document term matrix & sentimental analysis', 'summary': "Covers nlp concepts such as document term matrix and sentimental analysis, where a 73% accuracy is achieved using naive bias classifier for sentiment analysis in python, along with a discussion on a machine learning engineer master's program offered by edureka.", 'duration': 738.818, 'highlights': ['A 73% accuracy is achieved using a naive bias classifier for sentimental analysis in Python.', "The machine learning master's program at Edureka involves 200 plus hours of interactive training and covers nine modules.", 'The document term matrix is explained as a matrix with documents as rows and words as columns, used to understand the frequency of words in documents.', 'The chapter discusses the application of sentimental analysis, commonly used in marketing campaigns, social media, and e-commerce websites.', "The machine learning master's program at Edureka offers modules covering Python programming, machine learning, graphical modeling, reinforcement learning, NLP, AI and deep learning with TensorFlow, PySpark, and Python Spark using PySpark."]}], 'duration': 2856.362, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/JMUxmLyrhSk/pics/JMUxmLyrhSk14702161.jpg', 'highlights': ['TensorFlow training achieves mean squared error of 0.0029', '73% accuracy in sentimental analysis using naive bias classifier', 'Recurrent neural networks essential for predicting outputs based on past information', 'Importance of text mining and NLP in various applications', 'Importance of scaling and normalization of data for neural networks']}], 'highlights': ['Python is the most effective language for AI with its simplicity, extensive libraries, and widespread usage, making it the best choice for artificial intelligence over R, Java, Lisp, Prolog, and other languages.', 'The session covers the history, applications, and basics of artificial intelligence, including machine learning and deep learning concepts, with a practical implementation of NLP using Python.', 'Understanding the different types of neural networks, back propagation, and a deep learning demo will be covered in the session.', 'The different types of machine learning, algorithms involved, and demos showcasing real-world problem solving using machine learning will be discussed.', 'Real-World Applications of AI', "AI's Evolution from 1950s to Present", "Gmail uses AI and machine learning to classify emails as spam and non-spam based on words and correlations, such as 'lottery' and 'full refund', and separate them into different sections, demonstrating practical AI applications.", 'Supervised learning involves training the machine using well-labeled data, guiding the machine to understand patterns, and classifying input data into labeled output, essential for the training data set.', 'The speed limit variable provides the maximum information gain of 1 for car speed prediction', 'The chapter discusses the process of data splicing, involving the splitting of a data set into training and testing data, with 80% assigned for training and 20% for testing', 'Random forest basically builds multiple decision trees for more accurate predictions.', "K-nearest neighbor algorithm classifies new data points based on neighboring data points' features", 'Reinforcement learning crucial in machine learning, with applications in self-driving cars and deep learning such as AlphaGo', 'The agent performs 10,000 iterations during the training phase to find the best policy.', 'Deep learning automates feature extraction and mimics the functioning of the human brain through artificial neural networks', 'Back propagation is a supervised learning method for multilayer perceptrons, crucial for adjusting the weights during the training process of a neural network.', 'TensorFlow training achieves mean squared error of 0.0029']}