title
2020 Machine Learning Roadmap (95% valid for 2023)

description
Getting into machine learning is quite the adventure. And as any adventurer knows, sometimes it can be helpful to have a compass to figure out if you're heading in the right direction. Although the title of this video says machine learning roadmap, you should treat it as a compass. Explore it, follow your curiosity, learn something and use what you learn to create your next steps. Links: Interactive Machine Learning Roadmap - https://dbourke.link/mlmap Machine Learning Roadmap Resources - https://github.com/mrdbourke/machine-learning-roadmap Learn ML (beginner-friendly courses I teach) - https://www.mrdbourke.com/ml-courses/ ML courses/books I recommend - https://www.mrdbourke.com/ml-resources/ Read my novel Charlie Walks - https://www.charliewalks.com Timestamps: 0:00 - Hello & logistics 0:57 - PART 0: INTRO 1:42 - Brief overview of topics 3:05 - What is machine learning? 4:37 - Machine learning vs. traditional programming 7:41 - Why use machine learning? 8:44 - The number 1 rule of machine learning 10:45 - What is machine learning good for? 14:27 - How Tesla uses machine learning 17:57 - What we're going to cover in this video 20:52 - PART 1: Machine Learning Problems 22:27 - Categories of learning 26:17 - Machine learning problem domains 29:04 - Classification 33:57 - Regression 39:35 - PART 2: Machine Learning Process 41:57 - 6 major steps in a machine learning project 43:57 - Data collection 49:15 - Data preparation 1:04:00 - Training a model 1:23:33 - Analysis/evaluation 1:26:40 - Serving a model 1:29:09 - Retraining a model 1:30:07 - An example machine learning project 1:33:15 - PART 3: Machine Learning Tools 1:34:20 - Machine learning tools overview 1:38:36 - Machine learning toolbox (experiment tracking) 1:39:54 - Pretrained models for transfer learning 1:41:49 - Data and model tracking 1:43:35 - Cloud compute services 1:47:07 - Deep learning hardware (build your own deep learning PC) 1:47:53 - AutoML (automatic machine learning) 1:51:47 - Explainability (explaining the outputs of your machine learning model) 1:53:38 - Machine learning lifecycle (tools for end-to-end projects) 1:59:24 - PART 4: Machine Learning Mathematics 1:59:37 - The main branches of mathematics used in machine learning 2:03:16 - How I learn the math for machine learning 2:06:37 - PART 5: Machine Learning Resources 2:07:17 - A warning 2:08:42 - Where to start learning machine learning 2:14:51 - Made with ML (one of my favourite new websites for ML) 2:16:07 - Wokera ai (test your AI skills) 2:17:17 - A beginner-friendly path to start machine learning 2:19:02 - An advanced path for learning machine learning (after the beginner path) 2:21:43 - Where to learn the mathematics for machine learning 2:22:23 - Books for machine learning 2:24:27 - Where to learn cloud services 2:24:47 - Helpful rules and tidbits of machine learning 2:26:05 - How and why you should create your own blog 2:28:29 - Example machine learning curriculums 2:30:19 - Useful machine learning websites to visit 2:30:59 - Open-source datasets 2:31:26 - How to learn how to learn 2:32:57 - PART 6: Summary & Next Steps Connect elsewhere: Get email updates on my work - https://dbourke.link/newsletter Support on Patreon - https://bit.ly/mrdbourkepatreon Web - https://dbourke.link/web Quora - https://dbourke.link/quora Medium - https://dbourke.link/medium Twitter - https://dbourke.link/twitter LinkedIn - https://dbourke.link/linkedin #machinelearning #datascience

detail
{'title': '2020 Machine Learning Roadmap (95% valid for 2023)', 'heatmap': [{'end': 2453.321, 'start': 2355.219, 'weight': 0.891}, {'end': 4623.779, 'start': 4526.957, 'weight': 1}], 'summary': 'Presents a comprehensive machine learning roadmap, covering basics, fundamentals, metrics, data types, ml techniques, model deployment, tools, workflows, computer science, and learning resources, with practical guidance and emphasis on key concepts, tools, and learning paths.', 'chapters': [{'end': 393.702, 'segs': [{'end': 74.554, 'src': 'embed', 'start': 47.654, 'weight': 1, 'content': [{'end': 52.856, 'text': "And if there's anything else that's missing, if you have any questions, leave a comment below and I'll get back to you.", 'start': 47.654, 'duration': 5.202}, {'end': 55.713, 'text': 'But nonetheless, Enjoy.', 'start': 53.517, 'duration': 2.196}, {'end': 58.917, 'text': 'All right, all right, all right.', 'start': 55.733, 'duration': 3.184}, {'end': 70.509, 'text': 'Welcome to Machine Learning Roadmap for 2020, AKA a machine learning flavored visual interactive living mind map slash compass.', 'start': 59.838, 'duration': 10.671}, {'end': 74.554, 'text': "Well, that was a bit of a mouthful, but let's not spend any more time talking about it.", 'start': 71.03, 'duration': 3.524}], 'summary': 'Introduction to machine learning roadmap for 2020 with interactive living mind map.', 'duration': 26.9, 'max_score': 47.654, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk47654.jpg'}, {'end': 134.236, 'src': 'embed', 'start': 95.163, 'weight': 2, 'content': [{'end': 100.005, 'text': "I mean, whole textbooks have been written on machine learning, of course, but that's not what we're here for.", 'start': 95.163, 'duration': 4.842}, {'end': 106.026, 'text': "We're going to go through some of the main topics of machine learning, such as machine learning problems, the process,", 'start': 100.365, 'duration': 5.661}, {'end': 114.008, 'text': 'aka the steps in a machine learning project, the resources like how you might want to learn machine learning, the places you want to visit,', 'start': 106.026, 'duration': 7.982}, {'end': 122.23, 'text': 'the tools you can use to get the job done and, after all, since machine learning is basically mathematics under the hood,', 'start': 114.008, 'duration': 8.222}, {'end': 128.172, 'text': "we'll see what some of the main topics are in terms of what kinds of math runs our machine learning algorithms.", 'start': 122.23, 'duration': 5.942}, {'end': 131.374, 'text': "But how we're gonna do this, we're gonna be playful.", 'start': 128.792, 'duration': 2.582}, {'end': 134.236, 'text': "And that's what I want you to do with this resource here.", 'start': 131.894, 'duration': 2.342}], 'summary': 'Overview of machine learning topics, resources, and mathematics in a playful manner.', 'duration': 39.073, 'max_score': 95.163, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk95163.jpg'}, {'end': 289.81, 'src': 'embed', 'start': 196.64, 'weight': 0, 'content': [{'end': 200.702, 'text': 'And so for the sake of this video, I mean you could Google machine learning.', 'start': 196.64, 'duration': 4.062}, {'end': 202.643, 'text': 'you could get hundreds of different definitions.', 'start': 200.702, 'duration': 1.941}, {'end': 210.157, 'text': "but for the sake of this video, to keep it nice and simple, We're gonna treat machine learning as turning things,", 'start': 202.643, 'duration': 7.514}, {'end': 215.843, 'text': 'AKA data into numbers and finding patterns in those numbers.', 'start': 210.157, 'duration': 5.686}, {'end': 223.671, 'text': 'You might be wondering, well, how do you find those patterns in numbers? Well, the computer does this part.', 'start': 216.544, 'duration': 7.127}, {'end': 225.853, 'text': 'How? Math.', 'start': 224.352, 'duration': 1.501}, {'end': 229.471, 'text': "And again, we'll cover a little bit on this later.", 'start': 226.59, 'duration': 2.881}, {'end': 233.433, 'text': 'Now, if you want another one-liner definition of machine learning.', 'start': 230.072, 'duration': 3.361}, {'end': 240.156, 'text': 'machine learning is the field of study that gives computers the ability to learn without being explicitly programmed.', 'start': 233.433, 'duration': 6.723}, {'end': 241.797, 'text': 'Now that was by Arthur Samuel.', 'start': 240.436, 'duration': 1.361}, {'end': 244.478, 'text': 'I think that was almost over 50 years ago now.', 'start': 242.197, 'duration': 2.281}, {'end': 249.48, 'text': "So that's a key point there, without being explicitly programmed.", 'start': 245.478, 'duration': 4.002}, {'end': 250.921, 'text': "So let's jump into.", 'start': 249.94, 'duration': 0.981}, {'end': 258.37, 'text': 'traditional programming, which you might call software 1.0, versus machine learning, software 2.0.', 'start': 252.028, 'duration': 6.342}, {'end': 261.613, 'text': "Now you might be wondering what's the difference between the two?", 'start': 258.37, 'duration': 3.243}, {'end': 265.475, 'text': "What's the difference between traditional programming, the difference between machine learning?", 'start': 261.673, 'duration': 3.802}, {'end': 274.901, 'text': 'Well, before we even get into the difference, I just want to let you know is that Machine learning, to put it into practice,', 'start': 266.036, 'duration': 8.865}, {'end': 277.923, 'text': 'requires traditional programming to exist.', 'start': 274.901, 'duration': 3.022}, {'end': 283.006, 'text': 'Whereas traditional programming does not require machine learning.', 'start': 278.503, 'duration': 4.503}, {'end': 289.81, 'text': 'So although machine learning is amazing, remember you will need some traditional programming skills to be able to use it.', 'start': 283.406, 'duration': 6.404}], 'summary': 'Machine learning turns data into numbers and finds patterns using math, enabling computers to learn without explicit programming.', 'duration': 93.17, 'max_score': 196.64, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk196640.jpg'}, {'end': 336.621, 'src': 'embed', 'start': 309.265, 'weight': 5, 'content': [{'end': 312.007, 'text': "we're in the intro part just now, so software 2.0.", 'start': 309.265, 'duration': 2.742}, {'end': 314.098, 'text': 'hmm, Click on the link.', 'start': 312.007, 'duration': 2.091}, {'end': 316.082, 'text': 'Oh, we got a blog post here.', 'start': 314.88, 'duration': 1.202}, {'end': 320.935, 'text': 'I sometimes see people refer to neural networks as just another tool in your machine learning toolbox.', 'start': 316.624, 'duration': 4.311}, {'end': 323.95, 'text': 'Neural networks are not just another classifier.', 'start': 322.168, 'duration': 1.782}, {'end': 328.334, 'text': 'They represent the beginning of a fundamental shift in how we write software.', 'start': 324.37, 'duration': 3.964}, {'end': 331.016, 'text': 'They are software 2.0.', 'start': 328.674, 'duration': 2.342}, {'end': 332.177, 'text': "So I'll let you read that.", 'start': 331.016, 'duration': 1.161}, {'end': 336.621, 'text': "I'm not gonna read through it all, but this is the kind of way that you can explore this roadmap.", 'start': 332.197, 'duration': 4.424}], 'summary': 'Neural networks signify the shift to software 2.0, not just another tool in machine learning.', 'duration': 27.356, 'max_score': 309.265, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk309265.jpg'}], 'start': 1.932, 'title': 'Machine learning basics', 'summary': 'Covers the basics of machine learning, defining it as turning data into numbers, finding patterns through math, and distinguishing traditional programming from machine learning. it emphasizes the necessity of traditional programming skills for using machine learning.', 'chapters': [{'end': 170.543, 'start': 1.932, 'title': 'Machine learning roadmap 2020', 'summary': 'Introduces a machine learning roadmap for 2020, emphasizing an interactive living mind map and compass, providing links to resources and emphasizing a playful exploration of machine learning topics.', 'duration': 168.611, 'highlights': ['The chapter introduces a machine learning roadmap for 2020, emphasizing an interactive living mind map and compass.', 'The chapter provides links to resources and emphasizes a playful exploration of machine learning topics.', 'The chapter encourages the audience to explore the field of machine learning through various topics such as machine learning problems, the process, resources, tools, and mathematics.']}, {'end': 393.702, 'start': 171.464, 'title': 'Understanding machine learning basics', 'summary': 'Covers the basics of machine learning, defining it as turning data into numbers, finding patterns through math, and distinguishing traditional programming (software 1.0) from machine learning (software 2.0), highlighting the necessity of traditional programming skills for using machine learning.', 'duration': 222.238, 'highlights': ['Machine learning is defined as turning things, AKA data, into numbers and finding patterns in those numbers.', 'Machine learning is the field of study that gives computers the ability to learn without being explicitly programmed.', 'Machine learning, to put it into practice, requires traditional programming to exist, while traditional programming does not require machine learning.', 'Neural networks represent the beginning of a fundamental shift in how we write software, referred to as software 2.0.', 'Traditional programming (software 1.0) is distinguished from machine learning (software 2.0), emphasizing the necessity of traditional programming skills for using machine learning.']}], 'duration': 391.77, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk1932.jpg', 'highlights': ['Machine learning is defined as turning things, AKA data, into numbers and finding patterns in those numbers.', 'The chapter introduces a machine learning roadmap for 2020, emphasizing an interactive living mind map and compass.', 'The chapter provides links to resources and emphasizes a playful exploration of machine learning topics.', 'The chapter encourages the audience to explore the field of machine learning through various topics such as machine learning problems, the process, resources, tools, and mathematics.', 'Machine learning is the field of study that gives computers the ability to learn without being explicitly programmed.', 'Neural networks represent the beginning of a fundamental shift in how we write software, referred to as software 2.0.', 'Machine learning, to put it into practice, requires traditional programming to exist, while traditional programming does not require machine learning.', 'Traditional programming (software 1.0) is distinguished from machine learning (software 2.0), emphasizing the necessity of traditional programming skills for using machine learning.']}, {'end': 2171.09, 'segs': [{'end': 446.94, 'src': 'embed', 'start': 419.01, 'weight': 0, 'content': [{'end': 425.092, 'text': "And then it's gonna figure out the instructions to what the recipe is.", 'start': 419.01, 'duration': 6.082}, {'end': 433.534, 'text': 'So rather than us explicitly writing these instructions, this is the machine learning algorithm figuring out patterns in data.', 'start': 425.572, 'duration': 7.962}, {'end': 442.117, 'text': 'Actually, before it could even figure out these patterns, it would have to figure out some way to translate these inputs and outputs into numbers.', 'start': 433.774, 'duration': 8.343}, {'end': 446.94, 'text': "And how you do that is going to depend on what problem you're working on.", 'start': 443.439, 'duration': 3.501}], 'summary': 'Machine learning algorithm figures out recipe instructions and patterns in data.', 'duration': 27.93, 'max_score': 419.01, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk419010.jpg'}, {'end': 656.376, 'src': 'embed', 'start': 629.036, 'weight': 1, 'content': [{'end': 633.68, 'text': "Before machine learning, rule number one, don't be afraid to launch a product without machine learning.", 'start': 629.036, 'duration': 4.644}, {'end': 638.404, 'text': 'As we talked about before, software 1.0, aka hand coding.', 'start': 633.82, 'duration': 4.584}, {'end': 645.189, 'text': "everything can exist without machine learning, but machine learning can't exist without software 1.0..", 'start': 638.404, 'duration': 6.785}, {'end': 647.391, 'text': 'So what is machine learning good for?', 'start': 645.189, 'duration': 2.202}, {'end': 653.196, 'text': 'And I just realized that that should probably be what is machine learning good for, rather than this incorrect grammar.', 'start': 647.791, 'duration': 5.405}, {'end': 656.376, 'text': "Anyway, it doesn't matter.", 'start': 655.276, 'duration': 1.1}], 'summary': 'Launch product without machine learning, as it can exist without it. machine learning is dependent on software 1.0.', 'duration': 27.34, 'max_score': 629.036, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk629036.jpg'}, {'end': 1402.265, 'src': 'embed', 'start': 1374.284, 'weight': 2, 'content': [{'end': 1379.667, 'text': "And what I've done is I've correlated, as I said before, the sections here to the sections here.", 'start': 1374.284, 'duration': 5.383}, {'end': 1382.708, 'text': 'So if we come in here, this is how I want you to use this roadmap.', 'start': 1380.347, 'duration': 2.361}, {'end': 1388.152, 'text': "We've got machine learning problems, We break it out, we come up here, categories, types of learning.", 'start': 1382.728, 'duration': 5.424}, {'end': 1391.735, 'text': 'So supervised learning, what is that? You have data and you have labels.', 'start': 1388.352, 'duration': 3.383}, {'end': 1398.622, 'text': 'In our self-driving car example, our data might be images of tunnels, images of roads, images of people crossing the road.', 'start': 1392.156, 'duration': 6.466}, {'end': 1402.265, 'text': 'And the labels might be what those things actually are.', 'start': 1399.122, 'duration': 3.143}], 'summary': 'Correlated sections, roadmap for machine learning problems, self-driving car example with image data and labels.', 'duration': 27.981, 'max_score': 1374.284, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk1374284.jpg'}, {'end': 1573.273, 'src': 'embed', 'start': 1535.373, 'weight': 3, 'content': [{'end': 1537.834, 'text': 'Now, transfer learning is a very valuable skill.', 'start': 1535.373, 'duration': 2.461}, {'end': 1545.515, 'text': 'So we got here transfer learning because oftentimes, in practice, training a machine learning model.', 'start': 1539.451, 'duration': 6.064}, {'end': 1547.236, 'text': "so say you're building a self-driving car.", 'start': 1545.515, 'duration': 1.721}, {'end': 1555.122, 'text': 'for example, from that Tesla autonomy day video that we talked about, that took 70, 000 hours to train on a GPU cluster or GPU hours.', 'start': 1547.236, 'duration': 7.886}, {'end': 1556.723, 'text': 'Now you might not have access to that.', 'start': 1555.422, 'duration': 1.301}, {'end': 1562.787, 'text': 'but the beautiful thing about transfer learning is that you can take what another machine learning model has,', 'start': 1556.723, 'duration': 6.064}, {'end': 1573.273, 'text': 'AKA the patterns a machine learning model has learned on a particular data set, adjust it to your own and then use it for your own problem.', 'start': 1562.787, 'duration': 10.486}], 'summary': 'Transfer learning can save time: e.g. 70,000 hours for self-driving car model.', 'duration': 37.9, 'max_score': 1535.373, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk1535373.jpg'}, {'end': 1953.833, 'src': 'embed', 'start': 1916.04, 'weight': 4, 'content': [{'end': 1918.762, 'text': 'Is this video about machine learning on its own??', 'start': 1916.04, 'duration': 2.722}, {'end': 1923.506, 'text': 'Or is it about machine learning problems? Or is it about machine learning resources??', 'start': 1919.543, 'duration': 3.963}, {'end': 1925.608, 'text': 'It could actually be about all of them.', 'start': 1924.027, 'duration': 1.581}, {'end': 1927.41, 'text': "So therefore it's multi-label.", 'start': 1926.069, 'duration': 1.341}, {'end': 1932.694, 'text': 'Now, if we come back to classification, remember classification is a type of machine learning problem.', 'start': 1928.15, 'duration': 4.544}, {'end': 1934.736, 'text': "And then there's example problems.", 'start': 1933.435, 'duration': 1.301}, {'end': 1937.178, 'text': "And then we've got evaluation metrics.", 'start': 1935.477, 'duration': 1.701}, {'end': 1945.567, 'text': 'In other words, how do we know how well our machine learning model has learned different patterns for different problem sets?', 'start': 1938.021, 'duration': 7.546}, {'end': 1948.569, 'text': 'So then we can evaluate it with a confusion matrix.', 'start': 1946.167, 'duration': 2.402}, {'end': 1953.833, 'text': "And you might be wondering, what's a confusion matrix? I'm a little bit confused about a confusion matrix.", 'start': 1948.949, 'duration': 4.884}], 'summary': 'Discussion on machine learning, including classification and evaluation metrics.', 'duration': 37.793, 'max_score': 1916.04, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk1916040.jpg'}], 'start': 393.702, 'title': 'Machine learning fundamentals', 'summary': 'Explains the process and reasons for using machine learning, discusses fundamental concepts, such as key rules and practical applications, and delves into the machine learning roadmap, types of machine learning problems, and evaluation metrics, providing a comprehensive understanding of the subject.', 'chapters': [{'end': 465.484, 'start': 393.702, 'title': 'Machine learning basics', 'summary': 'Explains the process of machine learning, which involves a machine learning algorithm analyzing a set of inputs and ideal outputs to figure out patterns in data, with the goal of automating the process of instruction writing. it also addresses the question of why machine learning is used.', 'duration': 71.782, 'highlights': ['Machine learning algorithm starts with inputs and ideal outputs to figure out patterns in data', 'Goal: Automating the process of instruction writing', 'Reason for using machine learning']}, {'end': 1169.859, 'start': 466.471, 'title': 'Machine learning basics', 'summary': "Discusses the fundamental concepts of machine learning, including the reasons for using machine learning, the key rules for machine learning, and practical applications such as tesla's use of machine learning in self-driving cars.", 'duration': 703.388, 'highlights': ['Machine learning is good for problems with long lists of rules, continually changing environments, and discovering insights within large collections of data.', "Google's machine learning crash course offers 43 rules, emphasizing the importance of not being afraid to launch a product without machine learning and the suitability of machine learning for problems with long lists of rules, continually changing environments, and discovering insights within large collections of data.", "Tesla's autonomy day video showcases how machine learning is utilized in the production of self-driving cars, highlighting the use of machine learning algorithms to interpret data from car sensors and cameras to navigate complex driving environments."]}, {'end': 1485.661, 'start': 1170.279, 'title': 'Machine learning roadmap', 'summary': 'Discusses the approach to the machine learning roadmap, emphasizing exploration, feedback, and problem-solving, along with an overview of machine learning problems, categories of learning, and the use of the roadmap for understanding supervised learning.', 'duration': 315.382, 'highlights': ['The chapter emphasizes the approach to the machine learning roadmap, focusing on exploration, feedback, and problem-solving.', 'The chapter provides an overview of machine learning problems, including categories of learning such as supervised learning, unsupervised learning, and transfer learning.', 'The chapter explains the concept of supervised learning, highlighting the relationship between data and labels and providing examples such as self-driving cars and cooking.']}, {'end': 2171.09, 'start': 1486.622, 'title': 'Types of machine learning problems', 'summary': 'Provides an overview of unsupervised learning, reinforcement learning, and transfer learning, with examples and applications, and emphasizes the importance of transfer learning in reducing training time and effort, followed by a detailed explanation of classification, regression, clustering, and dimensionality reduction as common machine learning problems, along with evaluation metrics such as confusion matrix and example problems for each type.', 'duration': 684.468, 'highlights': ['Transfer learning is a valuable skill, saving significant training time, such as the 70,000 hours taken to train a self-driving car model, and enabling the adaptation of patterns learned by one model to solve another problem.', 'Examples of classification problems include binary classification (e.g., spam detection), multi-class classification (e.g., traffic light color detection), and multi-label classification (e.g., identifying multiple topics in a video).', 'Regression examples involve predicting the sale price of a house based on features like the number of bedrooms, bathrooms, and location, and the difficulty of predicting phenomena like Bitcoin prices is highlighted.', 'The concept and application of dimensionality reduction are explained, emphasizing the importance of reducing data to extract the most important features and avoid the curse of dimensionality.', 'Evaluation metrics such as confusion matrix are introduced, providing a method to evaluate the performance of machine learning models and compare actual labels to predicted labels.']}], 'duration': 1777.388, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk393702.jpg', 'highlights': ['Machine learning algorithm starts with inputs and ideal outputs to figure out patterns in data', "Google's machine learning crash course offers 43 rules, emphasizing the importance of not being afraid to launch a product without machine learning", 'The chapter emphasizes the approach to the machine learning roadmap, focusing on exploration, feedback, and problem-solving', 'Transfer learning is a valuable skill, saving significant training time, such as the 70,000 hours taken to train a self-driving car model', 'Examples of classification problems include binary classification (e.g., spam detection), multi-class classification (e.g., traffic light color detection), and multi-label classification (e.g., identifying multiple topics in a video)', 'Evaluation metrics such as confusion matrix are introduced, providing a method to evaluate the performance of machine learning models and compare actual labels to predicted labels']}, {'end': 2759.868, 'segs': [{'end': 2233.237, 'src': 'embed', 'start': 2171.671, 'weight': 0, 'content': [{'end': 2175.072, 'text': "I know I probably couldn't, so that's where you'd wanna probably use machine learning.", 'start': 2171.671, 'duration': 3.401}, {'end': 2176.312, 'text': 'We come back in.', 'start': 2175.612, 'duration': 0.7}, {'end': 2179.394, 'text': 'To evaluate how your machine learning model is going there.', 'start': 2176.793, 'duration': 2.601}, {'end': 2185.618, 'text': 'in a regression problem, you usually want to use R squared or mean squared error or mean absolute error.', 'start': 2179.394, 'duration': 6.224}, {'end': 2190.741, 'text': 'So mean absolute error is all errors are on the same scale, e.g.', 'start': 2186.318, 'duration': 4.423}, {'end': 2192.362, 'text': 'if trying to predict 100.', 'start': 2190.861, 'duration': 1.501}, {'end': 2196.343, 'text': 'Predicting 99 is the same error as predicting 101.', 'start': 2192.362, 'duration': 3.981}, {'end': 2199.966, 'text': 'Whereas squared error, it makes the outliers stand out more.', 'start': 2196.343, 'duration': 3.623}, {'end': 2208.594, 'text': 'That means if being 10% off is more than twice as bad as being 5% off, you probably want to pay more attention to mean squared error.', 'start': 2200.387, 'duration': 8.207}, {'end': 2210.515, 'text': "Again, we've got some links here.", 'start': 2208.894, 'duration': 1.621}, {'end': 2214.659, 'text': "We're blistering through this, aren't we?", 'start': 2213.258, 'duration': 1.401}, {'end': 2221.105, 'text': "Now, another problem that we haven't mentioned here on the machine learning problems is sequence to sequence.", 'start': 2215.019, 'duration': 6.086}, {'end': 2228.555, 'text': 'So if we got here, sequence to sequence is taking a sequence of something and turning it into a sequence of something else.', 'start': 2222.472, 'duration': 6.083}, {'end': 2233.237, 'text': 'In other words, given a sequence of English text translated into French.', 'start': 2229.555, 'duration': 3.682}], 'summary': 'Using machine learning for regression evaluation with r squared, mean squared error, and mean absolute error. also considering sequence to sequence translation.', 'duration': 61.566, 'max_score': 2171.671, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2171671.jpg'}, {'end': 2453.321, 'src': 'heatmap', 'start': 2355.219, 'weight': 0.891, 'content': [{'end': 2358.522, 'text': "there's some of our main machine learning problems.", 'start': 2355.219, 'duration': 3.303}, {'end': 2361.266, 'text': "now, If we come back, we've done one branch.", 'start': 2358.522, 'duration': 2.744}, {'end': 2362.126, 'text': "That's beautiful.", 'start': 2361.466, 'duration': 0.66}, {'end': 2364.307, 'text': "Let's come back to our little keynote here.", 'start': 2362.666, 'duration': 1.641}, {'end': 2368.489, 'text': 'What are we up to next? Machine learning process.', 'start': 2364.507, 'duration': 3.982}, {'end': 2370.99, 'text': "All right, let's dive in.", 'start': 2369.509, 'duration': 1.481}, {'end': 2379.574, 'text': 'Now, this is probably the biggest part of the entire roadmap.', 'start': 2376.232, 'duration': 3.342}, {'end': 2386.217, 'text': "And you can probably guess why, because the process of doing machine learning is relatively, I mean, let's just have a look.", 'start': 2380.034, 'duration': 6.183}, {'end': 2390.614, 'text': 'We got here steps in a machine learning project, 172.', 'start': 2386.991, 'duration': 3.623}, {'end': 2394.497, 'text': 'Whoa, look at that.', 'start': 2390.614, 'duration': 3.883}, {'end': 2397.779, 'text': 'If we zoom right out, here we go.', 'start': 2395.337, 'duration': 2.442}, {'end': 2399.901, 'text': "We've broken it all down.", 'start': 2398.76, 'duration': 1.141}, {'end': 2406.286, 'text': "Now I want to warn you again is that we're going through this as cooks, not chemists.", 'start': 2401.022, 'duration': 5.264}, {'end': 2408.387, 'text': 'So this might not be exact.', 'start': 2406.746, 'duration': 1.641}, {'end': 2410.869, 'text': "This is just what I've built out of my own experience.", 'start': 2408.447, 'duration': 2.422}, {'end': 2412.691, 'text': 'What I found is most helpful.', 'start': 2411.43, 'duration': 1.261}, {'end': 2416.238, 'text': 'It breaks down into a series of subtopics.', 'start': 2413.536, 'duration': 2.702}, {'end': 2417.519, 'text': "We've got data collection.", 'start': 2416.278, 'duration': 1.241}, {'end': 2424.345, 'text': "So if you want to find patterns in data, how do you collect that data in the first place? Then you've got data preparation.", 'start': 2417.8, 'duration': 6.545}, {'end': 2428.989, 'text': 'Remember, as we said, machine learning model works on finding patterns in numbers.', 'start': 2424.865, 'duration': 4.124}, {'end': 2437.476, 'text': "So how do you turn that data that you collect into numbers? Then we've got right down here, choosing an algorithm.", 'start': 2429.049, 'duration': 8.427}, {'end': 2443.656, 'text': "or actually before that, we've got train model on data, which can be broken down into three steps.", 'start': 2438.854, 'duration': 4.802}, {'end': 2448.179, 'text': "If you think about this, choose an algorithm or, once you've had a little bit of experience, if you're a beginner,", 'start': 2443.716, 'duration': 4.463}, {'end': 2453.321, 'text': 'you probably never heard of these steps, but usually it goes choose a certain type of machine learning algorithm.', 'start': 2448.179, 'duration': 5.142}], 'summary': 'Machine learning process involves 172 steps, including data collection, preparation, and algorithm selection.', 'duration': 98.102, 'max_score': 2355.219, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2355219.jpg'}, {'end': 2448.179, 'src': 'embed', 'start': 2416.278, 'weight': 1, 'content': [{'end': 2417.519, 'text': "We've got data collection.", 'start': 2416.278, 'duration': 1.241}, {'end': 2424.345, 'text': "So if you want to find patterns in data, how do you collect that data in the first place? Then you've got data preparation.", 'start': 2417.8, 'duration': 6.545}, {'end': 2428.989, 'text': 'Remember, as we said, machine learning model works on finding patterns in numbers.', 'start': 2424.865, 'duration': 4.124}, {'end': 2437.476, 'text': "So how do you turn that data that you collect into numbers? Then we've got right down here, choosing an algorithm.", 'start': 2429.049, 'duration': 8.427}, {'end': 2443.656, 'text': "or actually before that, we've got train model on data, which can be broken down into three steps.", 'start': 2438.854, 'duration': 4.802}, {'end': 2448.179, 'text': "If you think about this, choose an algorithm or, once you've had a little bit of experience, if you're a beginner,", 'start': 2443.716, 'duration': 4.463}], 'summary': 'Data collection, preparation, algorithm choice, and model training are key steps in machine learning.', 'duration': 31.901, 'max_score': 2416.278, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2416278.jpg'}, {'end': 2547.172, 'src': 'embed', 'start': 2523.335, 'weight': 3, 'content': [{'end': 2530.218, 'text': "Remember how I said I broke this down into like six different subtopics? Well, that's all in one little beautiful colorful chart here.", 'start': 2523.335, 'duration': 6.883}, {'end': 2531.64, 'text': "We've got data collection.", 'start': 2530.679, 'duration': 0.961}, {'end': 2533.781, 'text': "We'll ask questions like what data exists?", 'start': 2531.82, 'duration': 1.961}, {'end': 2534.902, 'text': 'Where can you get it?', 'start': 2534.202, 'duration': 0.7}, {'end': 2536.263, 'text': 'Is the data public?', 'start': 2535.322, 'duration': 0.941}, {'end': 2538.105, 'text': 'Are there privacy concerns?', 'start': 2536.864, 'duration': 1.241}, {'end': 2540.046, 'text': 'Is it structured or unstructured?', 'start': 2538.685, 'duration': 1.361}, {'end': 2542.828, 'text': 'And what I mean by structured or unstructured is imagine.', 'start': 2540.066, 'duration': 2.762}, {'end': 2545.45, 'text': 'structured is like an Excel spreadsheet.', 'start': 2542.828, 'duration': 2.622}, {'end': 2547.172, 'text': "You've got rows and columns.", 'start': 2545.831, 'duration': 1.341}], 'summary': 'Data collection includes identifying existing data, sources, privacy concerns, and structure.', 'duration': 23.837, 'max_score': 2523.335, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2523335.jpg'}], 'start': 2171.671, 'title': 'Machine learning metrics and sequence to sequence', 'summary': 'Discusses the evaluation metrics including r squared, mean squared error, and mean absolute error, and the sequence to sequence problem in machine learning, covering steps like data collection, preparation, algorithm selection, model training, analysis, evaluation, deployment, and retraining.', 'chapters': [{'end': 2214.659, 'start': 2171.671, 'title': 'Machine learning metrics', 'summary': 'Discusses the evaluation metrics for machine learning models, highlighting the use of r squared, mean squared error, and mean absolute error in regression problems, emphasizing the different scale and impact of errors on model evaluation.', 'duration': 42.988, 'highlights': ['R squared, mean squared error, and mean absolute error are commonly used evaluation metrics in regression problems.', 'Mean absolute error ensures all errors are on the same scale, making it suitable for cases where all errors should be treated equally.', 'Squared error emphasizes outliers, making it more sensitive to larger errors, which is beneficial when being slightly off has a significant impact.']}, {'end': 2759.868, 'start': 2215.019, 'title': 'Sequence to sequence in machine learning', 'summary': 'Discusses the sequence to sequence problem in machine learning, the machine learning process, and the various steps involved, such as data collection, data preparation, algorithm selection, model training, analysis and evaluation, model deployment, and model retraining.', 'duration': 544.849, 'highlights': ['The sequence to sequence problem in machine learning involves translating a sequence of text from one language to another, such as translating English text into French or Spanish using Google Translate.', 'The machine learning process involves various steps, including data collection, data preparation, algorithm selection, model training, analysis and evaluation, model deployment, and model retraining.', 'The data collection phase involves considerations such as the existence of data sources, privacy concerns, data public availability, storage options, and the types of data, including structured and unstructured data.']}], 'duration': 588.197, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2171671.jpg', 'highlights': ['R squared, mean squared error, and mean absolute error are commonly used evaluation metrics in regression problems.', 'The machine learning process involves various steps, including data collection, data preparation, algorithm selection, model training, analysis and evaluation, model deployment, and model retraining.', 'Mean absolute error ensures all errors are on the same scale, making it suitable for cases where all errors should be treated equally.', 'The data collection phase involves considerations such as the existence of data sources, privacy concerns, data public availability, storage options, and the types of data, including structured and unstructured data.', 'Squared error emphasizes outliers, making it more sensitive to larger errors, which is beneficial when being slightly off has a significant impact.', 'The sequence to sequence problem in machine learning involves translating a sequence of text from one language to another, such as translating English text into French or Spanish using Google Translate.']}, {'end': 3490.391, 'segs': [{'end': 2789.462, 'src': 'embed', 'start': 2760.839, 'weight': 0, 'content': [{'end': 2768.741, 'text': "is more than 56, 400, then you've got ordinal data, data with order, then you've got time series data, just like that Bitcoin example.", 'start': 2760.839, 'duration': 7.902}, {'end': 2778.623, 'text': 'This is a time series data because it goes over a time period, one week, one month, there we go.', 'start': 2770.121, 'duration': 8.502}, {'end': 2780.704, 'text': "So that's over a time period.", 'start': 2779.264, 'duration': 1.44}, {'end': 2787.2, 'text': "Come back, now you've got another type of data which we touched on before, unstructured data.", 'start': 2781.994, 'duration': 5.206}, {'end': 2789.462, 'text': 'So data with no rigid structure.', 'start': 2787.68, 'duration': 1.782}], 'summary': 'Data can be categorized into ordinal, time series, and unstructured types.', 'duration': 28.623, 'max_score': 2760.839, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2760839.jpg'}, {'end': 2931.797, 'src': 'embed', 'start': 2904.266, 'weight': 1, 'content': [{'end': 2908.869, 'text': 'How might this fit in? So if we put that back on the screen, wonderful.', 'start': 2904.266, 'duration': 4.603}, {'end': 2911.548, 'text': "So let's have a look at data collection.", 'start': 2910.165, 'duration': 1.383}, {'end': 2913.512, 'text': 'We might have a data source.', 'start': 2912.269, 'duration': 1.243}, {'end': 2917.086, 'text': 'Oh, and this is jumping ahead, but that is okay.', 'start': 2914.304, 'duration': 2.782}, {'end': 2918.807, 'text': 'We have a data source.', 'start': 2917.787, 'duration': 1.02}, {'end': 2922.931, 'text': 'So the cars, as I said, these all have cameras on them.', 'start': 2919.048, 'duration': 3.883}, {'end': 2927.314, 'text': "So they'll be collecting data from the environment, taking photos, et cetera, et cetera.", 'start': 2923.031, 'duration': 4.283}, {'end': 2931.797, 'text': "That's gonna be stored in Tesla's server somewhere.", 'start': 2928.075, 'duration': 3.722}], 'summary': "Tesla cars with cameras collect data for storage in tesla's servers.", 'duration': 27.531, 'max_score': 2904.266, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2904266.jpg'}, {'end': 3204.732, 'src': 'embed', 'start': 3180.406, 'weight': 5, 'content': [{'end': 3187.211, 'text': 'What are the outliers? How many of them are there? Are they out by much? Three plus standard deviations.', 'start': 3180.406, 'duration': 6.805}, {'end': 3189.092, 'text': 'Now, this is just a little heuristic.', 'start': 3187.531, 'duration': 1.561}, {'end': 3190.053, 'text': "You don't have to use that.", 'start': 3189.172, 'duration': 0.881}, {'end': 3193.035, 'text': "An outlier will really depend on the data that you're working with.", 'start': 3190.153, 'duration': 2.882}, {'end': 3198.058, 'text': "One way to find outliers is to plot a histogram, plot a distribution of your data if you're not sure what that is.", 'start': 3193.615, 'duration': 4.443}, {'end': 3199.139, 'text': "We'll touch on that in a minute.", 'start': 3198.198, 'duration': 0.941}, {'end': 3201.55, 'text': 'Why are there outliers there?', 'start': 3200.089, 'duration': 1.461}, {'end': 3202.731, 'text': 'Are there questions?', 'start': 3202.11, 'duration': 0.621}, {'end': 3204.732, 'text': 'you could ask a domain expert about the data?', 'start': 3202.731, 'duration': 2.001}], 'summary': 'Identify outliers; consider three plus standard deviations. use histograms for clarity.', 'duration': 24.326, 'max_score': 3180.406, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3180406.jpg'}, {'end': 3248.736, 'src': 'embed', 'start': 3223.264, 'weight': 2, 'content': [{'end': 3230.687, 'text': 'Exploratory data analysis, the main idea of that is to become one with your data, to build your own machine learning model.', 'start': 3223.264, 'duration': 7.423}, {'end': 3234.829, 'text': 'Now data pre-processing, preparing your data to be modeled.', 'start': 3231.367, 'duration': 3.462}, {'end': 3238.767, 'text': "Now, we've got a few steps here.", 'start': 3237.205, 'duration': 1.562}, {'end': 3243.912, 'text': "Again, it's probably wise because we did cover EDA a bit.", 'start': 3238.787, 'duration': 5.125}, {'end': 3245.653, 'text': "Let's just jump to the major topics.", 'start': 3244.152, 'duration': 1.501}, {'end': 3248.736, 'text': 'Feature imputation, filling missing values.', 'start': 3246.134, 'duration': 2.602}], 'summary': 'Exploratory data analysis and data pre-processing are key steps in building a machine learning model.', 'duration': 25.472, 'max_score': 3223.264, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3223264.jpg'}, {'end': 3317.19, 'src': 'embed', 'start': 3288.658, 'weight': 4, 'content': [{'end': 3292.999, 'text': 'So once you filled your missing values in a dataset.', 'start': 3288.658, 'duration': 4.341}, {'end': 3294.76, 'text': 'you typically do this with structured data.', 'start': 3292.999, 'duration': 1.761}, {'end': 3299.742, 'text': 'by the way, feature encoding is turning your dataset into numbers.', 'start': 3294.76, 'duration': 4.982}, {'end': 3303.903, 'text': 'Now, machine learning model requires all values to be numerical.', 'start': 3300.382, 'duration': 3.521}, {'end': 3311.806, 'text': "no matter what you're trying to model whether it's text, whether it's sound waves, whether it's images it has to be in some form of numerical form.", 'start': 3303.903, 'duration': 7.903}, {'end': 3317.19, 'text': "then you've got one hot encoding, then you've got label encoder, embedding encoding.", 'start': 3312.466, 'duration': 4.724}], 'summary': 'Data preprocessing involves filling missing values and encoding features into numerical form for machine learning models.', 'duration': 28.532, 'max_score': 3288.658, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3288658.jpg'}, {'end': 3418, 'src': 'embed', 'start': 3385.245, 'weight': 3, 'content': [{'end': 3391.007, 'text': 'So this is where you input your own knowledge into your data.', 'start': 3385.245, 'duration': 5.762}, {'end': 3396.529, 'text': "So you might have a collection of data where you've just collected raw form from the environment.", 'start': 3391.608, 'duration': 4.921}, {'end': 3403.312, 'text': 'And now feature engineering is where you take your domain knowledge and encode that into the data.', 'start': 3397.189, 'duration': 6.123}, {'end': 3409.454, 'text': 'So transform the data into potentially more meaningful representations by adding domain knowledge.', 'start': 3403.892, 'duration': 5.562}, {'end': 3418, 'text': 'If we click this little article here, boom, Discover feature engineering, how to engineer features and how to get good at it.', 'start': 3409.914, 'duration': 8.086}], 'summary': 'Feature engineering encodes domain knowledge into data to create meaningful representations.', 'duration': 32.755, 'max_score': 3385.245, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3385245.jpg'}], 'start': 2760.839, 'title': 'Types of data and data collection process', 'summary': "Discusses types of data such as ordinal, time series, and unstructured data, and explains the data collection process using tesla's data engine as an example. it emphasizes the importance of exploratory data analysis.", 'chapters': [{'end': 3162.379, 'start': 2760.839, 'title': 'Types of data and data collection process', 'summary': "Discusses the types of data including ordinal, time series, and unstructured data, and explains the data collection process using tesla's data engine as an example, emphasizing the importance of exploratory data analysis.", 'duration': 401.54, 'highlights': ['Explaining different types of data such as ordinal, time series, and unstructured data.', "Using Tesla's data collection process as an example, highlighting the role of cameras in collecting data from the environment.", 'Emphasizing the importance of exploratory data analysis in understanding the data before applying machine learning algorithms.']}, {'end': 3490.391, 'start': 3162.939, 'title': 'Data analysis techniques', 'summary': 'Covers techniques for data analysis including handling missing values, identifying outliers, data preparation, feature imputation, feature encoding, and feature engineering.', 'duration': 327.452, 'highlights': ['Feature engineering involves encoding domain knowledge into the data to create more meaningful representations.', 'Handling missing values through feature imputation is essential for structured data, and feature encoding is necessary to convert non-numerical data into numerical form for machine learning models.', 'Identifying and addressing outliers is crucial, and techniques such as plotting histograms and consulting domain experts can aid in this process.']}], 'duration': 729.552, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk2760839.jpg', 'highlights': ['Explaining different types of data such as ordinal, time series, and unstructured data.', "Using Tesla's data collection process as an example, highlighting the role of cameras in collecting data from the environment.", 'Emphasizing the importance of exploratory data analysis in understanding the data before applying machine learning algorithms.', 'Feature engineering involves encoding domain knowledge into the data to create more meaningful representations.', 'Handling missing values through feature imputation is essential for structured data, and feature encoding is necessary to convert non-numerical data into numerical form for machine learning models.', 'Identifying and addressing outliers is crucial, and techniques such as plotting histograms and consulting domain experts can aid in this process.']}, {'end': 5185.923, 'segs': [{'end': 3573.604, 'src': 'embed', 'start': 3534.002, 'weight': 0, 'content': [{'end': 3540.245, 'text': "read through these things and just see if any of them relate to what you're working on or just you wanna learn more in general.", 'start': 3534.002, 'duration': 6.243}, {'end': 3540.746, 'text': 'hit the link.', 'start': 3540.245, 'duration': 0.501}, {'end': 3542.507, 'text': 'Feature selection.', 'start': 3541.806, 'duration': 0.701}, {'end': 3547.75, 'text': 'So selecting the most valuable features of your data set to model.', 'start': 3543.587, 'duration': 4.163}, {'end': 3552.052, 'text': 'Remember before when we removed those two features from trying to predict heart disease.', 'start': 3548.25, 'duration': 3.802}, {'end': 3554.073, 'text': 'The curse of dimensionality.', 'start': 3552.732, 'duration': 1.341}, {'end': 3555.374, 'text': "Let's have a look at that actually.", 'start': 3554.313, 'duration': 1.061}, {'end': 3557.635, 'text': 'Curse of dimensionality.', 'start': 3555.814, 'duration': 1.821}, {'end': 3560.017, 'text': 'What is this thing, I say?', 'start': 3558.636, 'duration': 1.381}, {'end': 3567.041, 'text': 'The curse of dimensionality refers to the various phenomena that arise when analyzing and organizing data in high dimensional spaces.', 'start': 3560.677, 'duration': 6.364}, {'end': 3569.421, 'text': 'that do not occur.', 'start': 3568.68, 'duration': 0.741}, {'end': 3573.604, 'text': 'So, basically, high dimensional spaces just means lots of different data,', 'start': 3569.481, 'duration': 4.123}], 'summary': 'Feature selection involves choosing valuable data features for modeling, like when removing features in predicting heart disease. the curse of dimensionality refers to issues in analyzing and organizing data in high dimensional spaces.', 'duration': 39.602, 'max_score': 3534.002, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3534002.jpg'}, {'end': 3873.45, 'src': 'embed', 'start': 3843.669, 'weight': 6, 'content': [{'end': 3845.29, 'text': 'How do we train a model?', 'start': 3843.669, 'duration': 1.621}, {'end': 3854.754, 'text': "Remember the three steps choose an algorithm, overfit the model so, meaning that it's actually learned too well, it's learned the patterns too well.", 'start': 3845.57, 'duration': 9.184}, {'end': 3856.915, 'text': 'reduce the overfitting with regularization.', 'start': 3854.754, 'duration': 2.161}, {'end': 3860.577, 'text': "Again, if you're a beginner to machine learning, you'll look at something like regularization.", 'start': 3856.975, 'duration': 3.602}, {'end': 3862.858, 'text': "you'll be like I've never heard of that word in my life.", 'start': 3860.577, 'duration': 2.281}, {'end': 3865.579, 'text': "Well, that's okay, we're gonna go through what it is.", 'start': 3863.178, 'duration': 2.401}, {'end': 3871.808, 'text': "Or another great trick is you just go, I'll show you how to use this.", 'start': 3867.043, 'duration': 4.765}, {'end': 3873.45, 'text': 'This is a search bar.', 'start': 3872.589, 'duration': 0.861}], 'summary': 'Training a model involves choosing an algorithm, overfitting, and reducing overfitting with regularization, especially for beginners.', 'duration': 29.781, 'max_score': 3843.669, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3843669.jpg'}, {'end': 4030.488, 'src': 'embed', 'start': 4002.073, 'weight': 3, 'content': [{'end': 4003.715, 'text': "That's a big part of machine learning right there.", 'start': 4002.073, 'duration': 1.642}, {'end': 4006.952, 'text': "So we'll come back to this, choosing the right estimator.", 'start': 4004.67, 'duration': 2.282}, {'end': 4013.436, 'text': 'In scikit-learn, which is a machine learning Python library, machine learning model is often referred to as estimator.', 'start': 4007.572, 'duration': 5.864}, {'end': 4020.421, 'text': "So if we go through this, let's say for our heart disease classification problem, we have start, we only have, let's say 300 patients.", 'start': 4013.516, 'duration': 6.905}, {'end': 4023.623, 'text': 'So we have yes, above 50 samples, predicting a category.', 'start': 4020.501, 'duration': 3.122}, {'end': 4025.304, 'text': "So we're doing classification.", 'start': 4024.024, 'duration': 1.28}, {'end': 4026.966, 'text': "So that's heart disease or not.", 'start': 4025.545, 'duration': 1.421}, {'end': 4028.927, 'text': 'Do you have labeled data? Yes.', 'start': 4027.306, 'duration': 1.621}, {'end': 4030.488, 'text': 'We have under 100K samples.', 'start': 4029.327, 'duration': 1.161}], 'summary': 'Using scikit-learn for heart disease classification with under 100k labeled samples.', 'duration': 28.415, 'max_score': 4002.073, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4002073.jpg'}, {'end': 4499.38, 'src': 'embed', 'start': 4477.295, 'weight': 7, 'content': [{'end': 4485.878, 'text': 'combine it with their 25 camera cars and train a completely new machine learning model in one big hit on all of the data that they have.', 'start': 4477.295, 'duration': 8.583}, {'end': 4491.013, 'text': "Now, again, I'm making that example up, but that's just how you can imagine these things.", 'start': 4487.15, 'duration': 3.863}, {'end': 4493.876, 'text': 'Batch learning, typically everything happens in one big go.', 'start': 4491.133, 'duration': 2.743}, {'end': 4497.799, 'text': 'Online learning, little by little in a constantly changing environment.', 'start': 4494.156, 'duration': 3.643}, {'end': 4499.38, 'text': 'Transfer learning.', 'start': 4498.64, 'duration': 0.74}], 'summary': 'Company combines 25 camera cars to train a new machine learning model in one big hit, illustrating batch and online learning, and transfer learning.', 'duration': 22.085, 'max_score': 4477.295, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4477295.jpg'}, {'end': 4623.779, 'src': 'heatmap', 'start': 4526.957, 'weight': 1, 'content': [{'end': 4534.582, 'text': 'So use the following resources for different transfer learning models TensorFlow Hub, PyTorch, Hub Hugging, Face Transformers, Detectron 2,', 'start': 4526.957, 'duration': 7.625}, {'end': 4535.583, 'text': 'to name a few.', 'start': 4534.582, 'duration': 1.001}, {'end': 4541.246, 'text': "Active learning, so that's where you get a machine learning model to figure out some things of its own.", 'start': 4536.263, 'duration': 4.983}, {'end': 4544.389, 'text': 'And you also correct it with a human in the loop.', 'start': 4541.647, 'duration': 2.742}, {'end': 4554.701, 'text': "So, for example, in the Tesla car scenario, our human in the loop may be finding the scenarios where the car doesn't perform very well,", 'start': 4545.009, 'duration': 9.692}, {'end': 4558.605, 'text': 'collecting more data on that and then putting it back into the machine learning model.', 'start': 4554.701, 'duration': 3.904}, {'end': 4561.033, 'text': 'just like that stop sign we talked about.', 'start': 4559.332, 'duration': 1.701}, {'end': 4564.435, 'text': 'Ensembling, not really a form of learning.', 'start': 4561.053, 'duration': 3.382}, {'end': 4566.857, 'text': "It's more combining different algorithms together.", 'start': 4564.715, 'duration': 2.142}, {'end': 4572.681, 'text': 'A random forest is an example of an ensemble machine learning algorithm.', 'start': 4566.877, 'duration': 5.804}, {'end': 4577.063, 'text': "So that kind of means you're leveraging the wisdom of the crowd.", 'start': 4573.081, 'duration': 3.982}, {'end': 4582.227, 'text': "So if you ask one person hey, what's the best direction to take, left or right??", 'start': 4577.684, 'duration': 4.543}, {'end': 4586.515, 'text': 'they might say right, but then, if you ask nine other people, they might say left.', 'start': 4583.174, 'duration': 3.341}, {'end': 4594.777, 'text': "So should you trust the nine people or should you trust the one person? Now, if we come here, we've got underfitting.", 'start': 4587.055, 'duration': 7.722}, {'end': 4599.818, 'text': "So underfitting happens when your model doesn't perform as well as you'd like on your data.", 'start': 4595.337, 'duration': 4.481}, {'end': 4606.219, 'text': "So that means that basically the model hasn't learned as much as your evaluation metric would like it to.", 'start': 4600.478, 'duration': 5.741}, {'end': 4610.22, 'text': 'And remember, if we come back up here, we go to our problems.', 'start': 4606.739, 'duration': 3.481}, {'end': 4613.97, 'text': "We've got a series of evaluation metrics.", 'start': 4611.408, 'duration': 2.562}, {'end': 4623.779, 'text': 'Say we wanted to train a classification model to 99.99% accurate, but our model is actually underfitting.', 'start': 4614.611, 'duration': 9.168}], 'summary': "Utilize tensorflow hub, pytorch, and active learning for transfer learning; ensembling combines algorithms for leveraging crowd wisdom; underfitting occurs when model doesn't meet performance expectations.", 'duration': 96.822, 'max_score': 4526.957, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4526957.jpg'}, {'end': 4740.541, 'src': 'embed', 'start': 4712.206, 'weight': 4, 'content': [{'end': 4715.448, 'text': 'You can fix this through the various regularization techniques.', 'start': 4712.206, 'duration': 3.242}, {'end': 4722.874, 'text': "Here we go here, we've got L1 lasso and L2 ridge regularization, dropout.", 'start': 4716.567, 'duration': 6.307}, {'end': 4724.977, 'text': 'So dropout is actually really convenient.', 'start': 4723.255, 'duration': 1.722}, {'end': 4731.725, 'text': "It basically says, hey, let's remove just random parts of our model so that the rest of it becomes better.", 'start': 4725.397, 'duration': 6.328}, {'end': 4735.917, 'text': 'pretty cool, right early stopping.', 'start': 4733.675, 'duration': 2.242}, {'end': 4740.541, 'text': 'so stop your model before the training of validation loss starts to increase much more.', 'start': 4735.917, 'duration': 4.624}], 'summary': 'Regularization techniques like l1, l2, dropout, and early stopping improve model performance.', 'duration': 28.335, 'max_score': 4712.206, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4712206.jpg'}, {'end': 4787.232, 'src': 'embed', 'start': 4758.895, 'weight': 2, 'content': [{'end': 4765.6, 'text': "what's important is that, as you start to have practice in working with different machine learning problems, you'll start to go okay,", 'start': 4758.895, 'duration': 6.705}, {'end': 4767.322, 'text': 'I know where dropout fits in now.', 'start': 4765.6, 'duration': 1.722}, {'end': 4769.583, 'text': 'I know where data augmentation fits in.', 'start': 4767.642, 'duration': 1.941}, {'end': 4772.285, 'text': 'I know where batch normalization fits in.', 'start': 4770.124, 'duration': 2.161}, {'end': 4774.387, 'text': 'All of this comes with practice.', 'start': 4773.166, 'duration': 1.221}, {'end': 4776.749, 'text': "What we're doing now is just tying things together.", 'start': 4774.427, 'duration': 2.322}, {'end': 4781.866, 'text': "And finally, once you've trained a model, you might do some hyperparameter tuning.", 'start': 4777.642, 'duration': 4.224}, {'end': 4787.232, 'text': 'So run a bunch of experiments with different model settings and see which works best.', 'start': 4782.407, 'duration': 4.825}], 'summary': 'Practice helps understand ml techniques. hyperparameter tuning improves model performance.', 'duration': 28.337, 'max_score': 4758.895, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4758895.jpg'}, {'end': 4946.094, 'src': 'embed', 'start': 4920.362, 'weight': 5, 'content': [{'end': 4925.064, 'text': 'In terms of improving our model, one way might be to increase the number of layers here.', 'start': 4920.362, 'duration': 4.702}, {'end': 4926.925, 'text': 'So increase the complexity of the model.', 'start': 4925.244, 'duration': 1.681}, {'end': 4929.206, 'text': 'Another way might be to decrease these.', 'start': 4927.425, 'duration': 1.781}, {'end': 4933.648, 'text': 'So you cut this off here and we only use six layers.', 'start': 4929.586, 'duration': 4.062}, {'end': 4936.41, 'text': 'So that is a form of hyperparameter tuning.', 'start': 4934.169, 'duration': 2.241}, {'end': 4943.053, 'text': 'Any setting that you change in a model by hand is a form of hyperparameter tuning.', 'start': 4938.29, 'duration': 4.763}, {'end': 4946.094, 'text': 'So we come here, this is what a research paper looks like.', 'start': 4943.673, 'duration': 2.421}], 'summary': 'Improving model by increasing layers and hyperparameter tuning', 'duration': 25.732, 'max_score': 4920.362, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk4920362.jpg'}], 'start': 3490.431, 'title': 'Ml techniques and practices', 'summary': 'Covers feature selection, curse of dimensionality, data preparation, model training, choosing the right estimator for heart disease classification, and regularization and hyperparameter tuning in machine learning, emphasizing the importance of experimentation, various techniques, and model explainability.', 'chapters': [{'end': 3632.771, 'start': 3490.431, 'title': 'Feature selection and curse of dimensionality', 'summary': 'Covers feature selection, curse of dimensionality, and various techniques like discretization, crossing and interaction features, and dimensionality reduction in machine learning, emphasizing the importance of experimentation and practice in understanding these concepts.', 'duration': 142.34, 'highlights': ['The curse of dimensionality refers to the various phenomena that arise when analyzing and organizing data in high dimensional spaces, which do not occur in low dimensional settings.', 'Feature selection involves selecting the most valuable features of a dataset to model, and it can include techniques like dimensionality reduction and feature importance analysis.', 'Emphasizes the importance of experimentation and practice in understanding machine learning concepts, encouraging readers to work through problems and projects to improve their understanding over time.']}, {'end': 4001.833, 'start': 3634.093, 'title': 'Data preparation and model training', 'summary': 'Covers data preparation including dealing with imbalances, data preprocessing in the case of twitter, data sourcing for tesla, and data splitting into training, validation, and test sets, as well as model training steps including choosing an algorithm, overfitting, and regularization.', 'duration': 367.74, 'highlights': ['Data splitting into training, validation, and test sets', 'Data preprocessing in the case of Twitter', 'Data sourcing and analysis for Tesla', 'Model training steps including overfitting and regularization', "Use of transfer learning and ResNet model in Tesla's scenario"]}, {'end': 4711.426, 'start': 4002.073, 'title': 'Choosing the right estimator for heart disease classification', 'summary': 'Outlines the process of choosing the right estimator for a heart disease classification problem, focusing on machine learning algorithms like random forest, neural networks, unsupervised learning, and learning strategies like batch learning, online learning, transfer learning, active learning, ensembling, underfitting, and overfitting.', 'duration': 709.353, 'highlights': ['The chapter outlines the process of choosing the right estimator for a heart disease classification problem', 'Focusing on machine learning algorithms like random forest, neural networks, unsupervised learning', 'Learning strategies like batch learning, online learning, transfer learning, active learning, ensembling, underfitting, and overfitting']}, {'end': 5185.923, 'start': 4712.206, 'title': 'Regularization and hyperparameter tuning in ml', 'summary': 'Discusses various regularization techniques such as l1 lasso, l2 ridge, dropout, early stopping, data augmentation, and batch normalization, and emphasizes the importance of hyperparameter tuning in improving machine learning models, with insights into the costs involved and the tools for model explainability.', 'duration': 473.717, 'highlights': ['The chapter discusses various regularization techniques such as L1 Lasso, L2 Ridge, dropout, early stopping, data augmentation, and batch normalization.', 'The chapter emphasizes the importance of hyperparameter tuning in improving machine learning models, with insights into the costs involved and the tools for model explainability.', 'The chapter explains the bias and variance trade-off in machine learning, highlighting the impact of high bias resulting in underfitting and high variance leading to overfitting.']}], 'duration': 1695.492, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk3490431.jpg', 'highlights': ['The curse of dimensionality refers to phenomena in high dimensional spaces (3)', 'Feature selection involves selecting valuable features, using techniques like dimensionality reduction (2)', 'Emphasizes the importance of experimentation and practice in understanding ML concepts (1)', 'The chapter outlines the process of choosing the right estimator for heart disease classification (4)', 'The chapter discusses various regularization techniques such as L1 Lasso, L2 Ridge, dropout (7)', 'The chapter emphasizes the importance of hyperparameter tuning in improving ML models (6)', 'Model training steps include overfitting and regularization (5)', 'Learning strategies like batch learning, online learning, transfer learning, active learning, ensembling (8)']}, {'end': 5759.809, 'segs': [{'end': 5212.737, 'src': 'embed', 'start': 5186.563, 'weight': 3, 'content': [{'end': 5190.444, 'text': 'So machine learning models are actually really good at finding patterns in numbers,', 'start': 5186.563, 'duration': 3.881}, {'end': 5196.026, 'text': "but sometimes they're so good they can find patterns in what is just random noise in data.", 'start': 5190.444, 'duration': 5.582}, {'end': 5199.287, 'text': "So that's some analysis slash evaluation.", 'start': 5197.186, 'duration': 2.101}, {'end': 5204.213, 'text': "Now finally, we've got serve model and deploying a model.", 'start': 5200.532, 'duration': 3.681}, {'end': 5206.594, 'text': "After you've been through all of this.", 'start': 5205.114, 'duration': 1.48}, {'end': 5212.737, 'text': "you've collected some data, you've prepared it, you've trained a model, you've done some analysis and evaluation.", 'start': 5206.594, 'duration': 6.143}], 'summary': 'Machine learning models can find patterns in data, including random noise.', 'duration': 26.174, 'max_score': 5186.563, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5186563.jpg'}, {'end': 5301.042, 'src': 'embed', 'start': 5270.834, 'weight': 0, 'content': [{'end': 5273.775, 'text': 'Now tools that you can use to do this again.', 'start': 5270.834, 'duration': 2.941}, {'end': 5279.456, 'text': "this space is rapidly evolving, so chances are, if you're watching this video, in a few months these might be outdated,", 'start': 5273.775, 'duration': 5.681}, {'end': 5282.337, 'text': 'but right now TensorFlow Serving is gonna be really good.', 'start': 5279.456, 'duration': 2.881}, {'end': 5285.438, 'text': "PyTorch Serving, Google's AI platform.", 'start': 5282.337, 'duration': 3.101}, {'end': 5293.52, 'text': 'you can make your model available as a REST API SageMaker, which is Amazon Web Services machine learning deployment tool.', 'start': 5285.438, 'duration': 8.082}, {'end': 5301.042, 'text': "And then you've got MLOps, which is kind of this thing, which is, if you've heard of software engineering, DevOps.", 'start': 5294.201, 'duration': 6.841}], 'summary': 'Various tools like tensorflow serving, pytorch serving, sagemaker, and mlops are available for deploying machine learning models as rest apis.', 'duration': 30.208, 'max_score': 5270.834, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5270834.jpg'}, {'end': 5418.863, 'src': 'embed', 'start': 5395.893, 'weight': 1, 'content': [{'end': 5403.598, 'text': 'Collect some data, prepare the data, train a model on your data, analyze and evaluate it, serve it, and then retrain a model.', 'start': 5395.893, 'duration': 7.705}, {'end': 5407, 'text': "If we come back, that's an example of how Tesla would do it.", 'start': 5404.058, 'duration': 2.942}, {'end': 5411.763, 'text': 'Now, I actually did this in one of my own projects about a month or so ago.', 'start': 5408.061, 'duration': 3.702}, {'end': 5412.904, 'text': 'I collected some data.', 'start': 5412.023, 'duration': 0.881}, {'end': 5416.543, 'text': 'from open images, which is a data source.', 'start': 5413.942, 'duration': 2.601}, {'end': 5418.863, 'text': 'I pre-process it, oh, actually, sorry.', 'start': 5417.183, 'duration': 1.68}], 'summary': "Process and train model on data from open images, similar to tesla's approach.", 'duration': 22.97, 'max_score': 5395.893, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5395893.jpg'}, {'end': 5488.748, 'src': 'embed', 'start': 5438.228, 'weight': 2, 'content': [{'end': 5439.428, 'text': "Don't worry, we're gonna get to tools.", 'start': 5438.228, 'duration': 1.2}, {'end': 5443.869, 'text': 'I analyzed and tracked my experiments with Weights and Biases dashboard.', 'start': 5440.328, 'duration': 3.541}, {'end': 5448.59, 'text': 'I made a user interface with Streamlit, which is a beautiful tool.', 'start': 5444.629, 'duration': 3.961}, {'end': 5456.631, 'text': 'I wrapped all of this in a Docker container, pushed the Docker container to GCR, and deployed my app with App Engine.', 'start': 5449.15, 'duration': 7.481}, {'end': 5466.013, 'text': "So this would be my machine learning ops, machine learning operations for replicating Airbnb's amenity detection with Detectron 2.", 'start': 5456.651, 'duration': 9.362}, {'end': 5471.957, 'text': "Now, retraining a model I didn't actually do, but if I was, I would find the worst performing classes.", 'start': 5466.013, 'duration': 5.944}, {'end': 5479.402, 'text': 'So this problem was using computer vision to detect amenities in photos.', 'start': 5472.577, 'duration': 6.825}, {'end': 5481.983, 'text': 'Amenity detection.', 'start': 5480.903, 'duration': 1.08}, {'end': 5487.347, 'text': 'This is what happened.', 'start': 5486.566, 'duration': 0.781}, {'end': 5488.748, 'text': 'This is the problem in a nutshell.', 'start': 5487.607, 'duration': 1.141}], 'summary': "Used tools like weights and biases, streamlit, docker, gcr, and app engine for ml ops in replicating airbnb's amenity detection with detectron 2.", 'duration': 50.52, 'max_score': 5438.228, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5438228.jpg'}, {'end': 5543.717, 'src': 'embed', 'start': 5516.172, 'weight': 5, 'content': [{'end': 5518.874, 'text': 'This was my process that I did.', 'start': 5516.172, 'duration': 2.702}, {'end': 5523.099, 'text': 'If I wanted to retrain a model, I would find the worst performing classes.', 'start': 5519.015, 'duration': 4.084}, {'end': 5532.108, 'text': 'So say my model sucked at finding chairs, I would get more photos of chairs, feed that back into my Google storage for storing all the models.', 'start': 5523.119, 'duration': 8.989}, {'end': 5533.289, 'text': 'Little typo there.', 'start': 5532.288, 'duration': 1.001}, {'end': 5537.714, 'text': 'and retrain a model and then go back through this process.', 'start': 5534.552, 'duration': 3.162}, {'end': 5543.717, 'text': "If you want to see how this was done, there's a link there, dburk.link slash Airbnb playlist.", 'start': 5537.734, 'duration': 5.983}], 'summary': 'Retraining model by adding more photos of worst performing classes and storing in google storage.', 'duration': 27.545, 'max_score': 5516.172, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5516172.jpg'}, {'end': 5647.637, 'src': 'embed', 'start': 5622.242, 'weight': 6, 'content': [{'end': 5630.487, 'text': 'Now, I say Python-flavored, but you can actually write machine learning code in any programming language that does or allows numerical computing.', 'start': 5622.242, 'duration': 8.245}, {'end': 5640.653, 'text': 'A lot of these libraries here, like scikit-learn, PyTorch, TensorFlow, and Python itself actually, execute C code under the hood, so really fast code.', 'start': 5631.127, 'duration': 9.526}, {'end': 5647.637, 'text': "But when you're first getting started, chances are you're gonna be interacting with one of these libraries, or one or more of these libraries.", 'start': 5641.033, 'duration': 6.604}], 'summary': 'Python and other languages can be used for machine learning, with fast execution using libraries like scikit-learn, pytorch, and tensorflow.', 'duration': 25.395, 'max_score': 5622.242, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5622242.jpg'}], 'start': 5186.563, 'title': "Ml model deployment and airbnb's amenity detection", 'summary': "Discusses challenges in deploying ml models, retraining models, and tools like tensorflow serving, pytorch serving, google's ai platform, sagemaker, and mlops. it also covers replicating airbnb's amenity detection, including retraining models, problem identification, and exploring ml tools and libraries.", 'chapters': [{'end': 5456.631, 'start': 5186.563, 'title': 'Machine learning model deployment', 'summary': "Discusses the challenges of deploying machine learning models, the importance of retraining models, and the tools and processes involved, such as tensorflow serving, pytorch serving, google's ai platform, sagemaker, and mlops.", 'duration': 270.068, 'highlights': ['Deploying machine learning models and the challenges involved in serving and retraining models are discussed, along with the importance of evaluation metrics and the real-world testing of models.', "Tools and processes for deploying machine learning models, including TensorFlow Serving, PyTorch Serving, Google's AI platform, SageMaker, and MLOps, are explored.", "The process of deploying a machine learning model is compared to Tesla's data engine, emphasizing the importance of collecting, preparing, training, analyzing, and retraining models in a loop.", 'The speaker shares his personal experience of deploying a machine learning model, highlighting the use of tools such as Python scripts, Google storage, Weights and Biases, Streamlit, and Docker container, along with the importance of tracking and analyzing experiments.']}, {'end': 5759.809, 'start': 5456.651, 'title': "Replicating airbnb's amenity detection", 'summary': "Covers replicating airbnb's amenity detection using machine learning ops and discusses the process of retraining a model, identifying the problem of amenity detection, and exploring machine learning tools and libraries.", 'duration': 303.158, 'highlights': ['The problem of amenity detection using computer vision in photos is discussed.', 'The process of retraining a model and finding worst performing classes is explained.', 'Exploration of machine learning tools and Python-flavored libraries for building machine learning models is discussed.']}], 'duration': 573.246, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5186563.jpg', 'highlights': ["Tools and processes for deploying ML models, including TensorFlow Serving, PyTorch Serving, Google's AI platform, SageMaker, and MLOps, are explored.", "The process of deploying a machine learning model is compared to Tesla's data engine, emphasizing the importance of collecting, preparing, training, analyzing, and retraining models in a loop.", 'The speaker shares his personal experience of deploying a machine learning model, highlighting the use of tools such as Python scripts, Google storage, Weights and Biases, Streamlit, and Docker container, along with the importance of tracking and analyzing experiments.', 'Deploying machine learning models and the challenges involved in serving and retraining models are discussed, along with the importance of evaluation metrics and the real-world testing of models.', 'The problem of amenity detection using computer vision in photos is discussed.', 'The process of retraining a model and finding worst performing classes is explained.', 'Exploration of machine learning tools and Python-flavored libraries for building machine learning models is discussed.']}, {'end': 6815.668, 'segs': [{'end': 5803.718, 'src': 'embed', 'start': 5760.409, 'weight': 0, 'content': [{'end': 5765.993, 'text': "So if you're writing code in TensorFlow and you need to track your experiments, you might use Dashboard by Weights and Biases.", 'start': 5760.409, 'duration': 5.584}, {'end': 5770.095, 'text': 'And if you want a pre-trained TensorFlow model, you might go to TensorFlow Hub.', 'start': 5766.573, 'duration': 3.522}, {'end': 5776.519, 'text': "And if you wanted to track the changes you're making to your data, you might use Weights and Biases artifacts.", 'start': 5770.575, 'duration': 5.944}, {'end': 5781.742, 'text': 'And if you want a cloud compute service, well, Google Colab is a free resource.', 'start': 5777.199, 'duration': 4.543}, {'end': 5787.626, 'text': 'But if you need a bigger amount of compute, you might use Google Cloud or AWS or Microsoft Azure.', 'start': 5782.383, 'duration': 5.243}, {'end': 5791.449, 'text': "And that is if you don't even have your own computing resources.", 'start': 5788.207, 'duration': 3.242}, {'end': 5796.933, 'text': "Now, truth be told, if you're first getting started, you're probably only going to be using Colab to begin with.", 'start': 5791.489, 'duration': 5.444}, {'end': 5799.074, 'text': "But then it's very helpful.", 'start': 5797.293, 'duration': 1.781}, {'end': 5803.718, 'text': 'I will recommend this in the learning resource to learn at least one cloud provider.', 'start': 5799.094, 'duration': 4.624}], 'summary': 'Tools for tensorflow: dashboard for experiments, tensorflow hub for pre-trained models, weights and biases for data tracking, google colab for free cloud compute, and options for larger compute needs with google cloud, aws, or microsoft azure. recommended to learn at least one cloud provider.', 'duration': 43.309, 'max_score': 5760.409, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5760409.jpg'}, {'end': 6064.666, 'src': 'embed', 'start': 6034.611, 'weight': 2, 'content': [{'end': 6039.453, 'text': 'How cool is that? Then the same thing is with PyTorch Hub.', 'start': 6034.611, 'duration': 4.842}, {'end': 6045.237, 'text': "Now I believe PyTorch Hub, it's not as well refined as TensorFlow Hub, but it's still got some pre-trained PyTorch models.", 'start': 6039.493, 'duration': 5.744}, {'end': 6046.677, 'text': 'Great offering there.', 'start': 6045.897, 'duration': 0.78}, {'end': 6048.378, 'text': 'Hugging Face Transformers.', 'start': 6047.158, 'duration': 1.22}, {'end': 6052.401, 'text': 'Now this is, Transformer is a natural language processing architecture.', 'start': 6048.438, 'duration': 3.963}, {'end': 6057.784, 'text': 'Now Hugging Face is a natural language processing research team.', 'start': 6054.322, 'duration': 3.462}, {'end': 6064.666, 'text': 'And they have the biggest open source repository of transformers.', 'start': 6058.74, 'duration': 5.926}], 'summary': 'Pytorch hub and hugging face offer pre-trained models for nlp.', 'duration': 30.055, 'max_score': 6034.611, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6034611.jpg'}, {'end': 6388.292, 'src': 'embed', 'start': 6363.635, 'weight': 4, 'content': [{'end': 6368.981, 'text': "It's formerly called Machine Learning Engine or Azure Machine Learning from Microsoft Azure.", 'start': 6363.635, 'duration': 5.346}, {'end': 6374.307, 'text': "Now these, okay, they're gonna have different names and fancy different marketing terms.", 'start': 6369.682, 'duration': 4.625}, {'end': 6380.93, 'text': "They're basically different versions of the same thing, depending on where you work or your own like.", 'start': 6374.368, 'duration': 6.562}, {'end': 6386.091, 'text': "personally, I just like Google Cloud Platform, because that's the one I've got most experience with.", 'start': 6380.93, 'duration': 5.161}, {'end': 6388.292, 'text': "I've used AWS in the past.", 'start': 6386.532, 'duration': 1.76}], 'summary': 'Microsoft azure has rebranded its machine learning engine, while the speaker prefers google cloud platform due to personal experience.', 'duration': 24.657, 'max_score': 6363.635, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6363635.jpg'}, {'end': 6512.145, 'src': 'embed', 'start': 6488.737, 'weight': 3, 'content': [{'end': 6496.124, 'text': 'Remember how I said right back at the start how machine learning is like figuring out or calculating the rules or figuring out the patterns in your data set.', 'start': 6488.737, 'duration': 7.387}, {'end': 6500.798, 'text': "Why not use machine learning just to build itself That's the premise of AutoML.", 'start': 6496.184, 'duration': 4.614}, {'end': 6503.119, 'text': 'You can also use that for hyper parameter tuning.', 'start': 6501.278, 'duration': 1.841}, {'end': 6512.145, 'text': "So, just like we were turning the dials on our oven when we're cooking our Sicilian grandmother's beautiful roast chicken dish,", 'start': 6503.64, 'duration': 8.505}], 'summary': 'Automl uses machine learning to build itself and for hyper parameter tuning.', 'duration': 23.408, 'max_score': 6488.737, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6488737.jpg'}], 'start': 5760.409, 'title': 'Machine learning tools and workflows', 'summary': 'Delves into various machine learning tools and resources, such as tensorflow, google colab, and automl, emphasizing the use of pre-trained models and the significance of tracking experiments. it also highlights the importance of workflow tips, including utilizing pre-trained models for transfer learning and the necessity of cloud compute services. additionally, it discusses the landscape of machine learning tools and platforms, focusing on cloud computing services and explainability tools to aid in building and deploying machine learning models.', 'chapters': [{'end': 5966.235, 'start': 5760.409, 'title': 'Machine learning tools and resources', 'summary': 'Discusses various tools and resources for machine learning, including tensorflow, google colab, cloud compute services, automl, and streamlit, emphasizing the importance of tracking experiments and using pre-trained models to build proof of concepts.', 'duration': 205.826, 'highlights': ['TensorFlow Hub and Weights and Biases are used to track experiments and changes to data for machine learning models, while Google Colab is recommended as a free resource (relevance score: 5)', 'Cloud compute services such as Google Cloud, AWS, and Microsoft Azure are essential for larger compute needs, with a recommendation to learn at least one cloud provider for serious machine learning work (relevance score: 4)', 'AutoML and hyperparameter tuning tools are valuable for improving model settings and building automatically generated machine learning models, with a focus on using transfer learning and AutoML rather than building models from scratch (relevance score: 3)', 'Machine learning lifecycle tools like Kubeflow, Selden, and MLflow are mentioned, with a note that these tools are typically used after at least a year of experience and when building machine learning powered applications or products (relevance score: 2)', 'Streamlit is recommended for building proof of concepts and deploying web dashboards for machine learning applications (relevance score: 1)']}, {'end': 6338.368, 'start': 5966.255, 'title': 'Machine learning workflow tips', 'summary': 'Highlights the importance of using pre-trained models for transfer learning, the significance of tracking changes to datasets and models using tools like weights and biases and dvc, and the necessity of cloud compute services, specifically gpus, for computationally intensive machine learning tasks.', 'duration': 372.113, 'highlights': ['Pre-trained models and transfer learning are essential in machine learning, with TensorFlow Hub and PyTorch Hub being key resources, and Hugging Face Transformers being crucial for text problems.', 'Tools like Weights and Biases and DVC enable tracking changes to datasets and models, ensuring reproducibility and facilitating version control.', 'The necessity of cloud compute services, particularly GPUs, for computationally intensive machine learning tasks, with Google Colab providing free access to Nvidia GPUs and TPUs.']}, {'end': 6815.668, 'start': 6338.768, 'title': 'Machine learning tools and platforms', 'summary': 'Discusses the landscape of machine learning tools and platforms, including cloud computing services like aws, google cloud platform, and microsoft azure, as well as automl tools like tpot and google cloud automl, and explainability tools like the what if tool and shap values, aiming to assist in building and deploying machine learning models.', 'duration': 476.9, 'highlights': ['The chapter discusses the landscape of machine learning tools and platforms, including cloud computing services like AWS, Google Cloud Platform, and Microsoft Azure.', 'AutoML tools like TPOT and Google Cloud AutoML are introduced, aiming to assist in building and deploying machine learning models.', 'The explainability tools like the What If Tool and SHAP values are discussed to help explain the outputs of machine learning models.']}], 'duration': 1055.259, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk5760409.jpg', 'highlights': ['Cloud compute services such as Google Cloud, AWS, and Microsoft Azure are essential for larger compute needs, with a recommendation to learn at least one cloud provider for serious machine learning work (relevance score: 4)', 'TensorFlow Hub and Weights and Biases are used to track experiments and changes to data for machine learning models, while Google Colab is recommended as a free resource (relevance score: 3)', 'Pre-trained models and transfer learning are essential in machine learning, with TensorFlow Hub and PyTorch Hub being key resources, and Hugging Face Transformers being crucial for text problems (relevance score: 2)', 'AutoML and hyperparameter tuning tools are valuable for improving model settings and building automatically generated machine learning models, with a focus on using transfer learning and AutoML rather than building models from scratch (relevance score: 1)', 'The chapter discusses the landscape of machine learning tools and platforms, including cloud computing services like AWS, Google Cloud Platform, and Microsoft Azure (relevance score: 0)']}, {'end': 7786.011, 'segs': [{'end': 6843.545, 'src': 'embed', 'start': 6817.93, 'weight': 6, 'content': [{'end': 6822.96, 'text': 'All right, now, machine learning lifecycle, this is probably gonna be a little bit further on in your journey.', 'start': 6817.93, 'duration': 5.03}, {'end': 6829.453, 'text': "When you're first starting out, you're gonna go through, I think, mostly just in this stage.", 'start': 6823.521, 'duration': 5.932}, {'end': 6835.458, 'text': "Like this is how I've actually tried to stage a lot of this roadmap, or compass, whatever you want to call it is.", 'start': 6829.493, 'duration': 5.965}, {'end': 6837.38, 'text': 'the branches here are in cascading effects.', 'start': 6835.458, 'duration': 1.922}, {'end': 6843.545, 'text': "So you'll start with the libraries, you'll track your experiments, you'll look for pre-trained models, or maybe you might do that before that.", 'start': 6837.4, 'duration': 6.145}], 'summary': 'Machine learning lifecycle involves stages like library selection, experiment tracking, and pre-trained model search.', 'duration': 25.615, 'max_score': 6817.93, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6817930.jpg'}, {'end': 6909.806, 'src': 'embed', 'start': 6852.998, 'weight': 7, 'content': [{'end': 6862.932, 'text': "But yeah, once we get into the full blown ML machine learning life cycle, you'll want to look at things like MLflow, Kubeflow, Selden, Streamlit.", 'start': 6852.998, 'duration': 9.934}, {'end': 6865.536, 'text': "Also, let's just have a look at one of these.", 'start': 6863.533, 'duration': 2.003}, {'end': 6868.44, 'text': 'Also that document that we just had a look at before.', 'start': 6866.077, 'duration': 2.363}, {'end': 6878.523, 'text': "We had a look at this one before, but this is actually, this article only came out a couple of days ago, so it's really worth checking out.", 'start': 6872.702, 'duration': 5.821}, {'end': 6879.643, 'text': 'I enjoyed it.', 'start': 6879.123, 'duration': 0.52}, {'end': 6883.104, 'text': "There's a great resource right at the top here.", 'start': 6880.143, 'duration': 2.961}, {'end': 6887.485, 'text': 'This is a guide to production level deep learning.', 'start': 6885.084, 'duration': 2.401}, {'end': 6890.386, 'text': "Maybe I'll just put that in there.", 'start': 6889.345, 'duration': 1.041}, {'end': 6901.2, 'text': "Boom And I encourage you actually, If this roadmap isn't supporting your needs, if you think it can be improved, just make your own.", 'start': 6892.706, 'duration': 8.494}, {'end': 6909.806, 'text': 'This will actually help you, what is it called? A guide to production level deep learning.', 'start': 6901.62, 'duration': 8.186}], 'summary': 'Explore ml life cycle tools like mlflow, kubeflow, selden, and streamlit. encouragement to create own guide to production level deep learning.', 'duration': 56.808, 'max_score': 6852.998, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6852998.jpg'}, {'end': 6972.556, 'src': 'embed', 'start': 6941.193, 'weight': 8, 'content': [{'end': 6944.014, 'text': 'MLflow tracking, projects, models, registry.', 'start': 6941.193, 'duration': 2.821}, {'end': 6946.735, 'text': 'Built-in integrations.', 'start': 6945.695, 'duration': 1.04}, {'end': 6947.696, 'text': 'These are all other.', 'start': 6946.795, 'duration': 0.901}, {'end': 6950.358, 'text': 'machine learning tools that you can use.', 'start': 6948.757, 'duration': 1.601}, {'end': 6955.843, 'text': "There we go, Azure Machine Learning, that's Microsoft's office, I mean offering, Microsoft Office.", 'start': 6950.919, 'duration': 4.924}, {'end': 6963.388, 'text': "SageMaker, we mentioned that, Google Cloud, Kubernetes, that's what Kubeflow is.", 'start': 6957.384, 'duration': 6.004}, {'end': 6967.312, 'text': "I imagine Kubeflow is just a workflow that's built on top of Kubernetes.", 'start': 6963.969, 'duration': 3.343}, {'end': 6972.556, 'text': "If you don't know what Kubernetes is, it's a framework for building containerized applications, e.g.", 'start': 6967.352, 'duration': 5.204}], 'summary': 'Mlflow supports integrations with azure ml, sagemaker, google cloud, and kubeflow, which is built on top of kubernetes.', 'duration': 31.363, 'max_score': 6941.193, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6941193.jpg'}, {'end': 7015.132, 'src': 'embed', 'start': 6987.383, 'weight': 5, 'content': [{'end': 6994.867, 'text': "these sort of tools won't be until later stage in your development, especially if you're just getting started with machine learning,", 'start': 6987.383, 'duration': 7.484}, {'end': 7001.31, 'text': "but they're worth knowing about, because, at the end of the day, if you want to build things with machine learning,", 'start': 6994.867, 'duration': 6.443}, {'end': 7002.631, 'text': "this is where you're going to end up.", 'start': 7001.31, 'duration': 1.321}, {'end': 7007.954, 'text': "If you want to build things that get into the hands of people, this is where you're going to end up.", 'start': 7002.911, 'duration': 5.043}, {'end': 7015.132, 'text': "One of the most beginner-friendly points here, actually, we'll put it up here, is Streamlit, or beginner-friendly tools, sorry.", 'start': 7009.349, 'duration': 5.783}], 'summary': 'Streamlit is a beginner-friendly tool for machine learning development.', 'duration': 27.749, 'max_score': 6987.383, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6987383.jpg'}, {'end': 7294.804, 'src': 'embed', 'start': 7266.236, 'weight': 10, 'content': [{'end': 7267.997, 'text': "If you went through high school, you might've covered them.", 'start': 7266.236, 'duration': 1.761}, {'end': 7273.08, 'text': 'Do you need to know the ins and outs of all of them to get started with machine learning?', 'start': 7268.537, 'duration': 4.543}, {'end': 7280.705, 'text': 'No, my approach is write some machine learning code and then learn these parts when you have to.', 'start': 7273.42, 'duration': 7.285}, {'end': 7282.606, 'text': "So let's come back.", 'start': 7281.885, 'duration': 0.721}, {'end': 7284.727, 'text': 'Machine learning mathematics.', 'start': 7283.687, 'duration': 1.04}, {'end': 7287.329, 'text': "What's running under the hood? There we go.", 'start': 7285.248, 'duration': 2.081}, {'end': 7290.58, 'text': 'So, linear algebra.', 'start': 7289.198, 'duration': 1.382}, {'end': 7294.804, 'text': 'Creating objects and a set of rules to manipulate these objects, e.g.', 'start': 7291.581, 'duration': 3.223}], 'summary': 'High school concepts not essential for starting machine learning; focus on practical code and learn as needed.', 'duration': 28.568, 'max_score': 7266.236, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7266236.jpg'}, {'end': 7522.598, 'src': 'embed', 'start': 7494.975, 'weight': 2, 'content': [{'end': 7501.178, 'text': "If you do wanna be an artificial intelligence researcher, chances are you're gonna have to know these inside out.", 'start': 7494.975, 'duration': 6.203}, {'end': 7504.06, 'text': 'So get good at math.', 'start': 7501.778, 'duration': 2.282}, {'end': 7508.945, 'text': 'read this book end to end the machine learning for mathematics book, my favorite resource for learning.', 'start': 7504.06, 'duration': 4.885}, {'end': 7510.106, 'text': 'mathematics for machine learning.', 'start': 7508.945, 'duration': 1.161}, {'end': 7510.947, 'text': 'Here you go.', 'start': 7510.486, 'duration': 0.461}, {'end': 7514.61, 'text': "Some of the things we've just talked about mathematical foundations.", 'start': 7511.087, 'duration': 3.523}, {'end': 7522.598, 'text': "You've also got the deep learning book, fast.ai deep learning from the foundations and various other resources.", 'start': 7515.992, 'duration': 6.606}], 'summary': "Aspiring ai researchers need strong math skills and can benefit from resources like the 'machine learning for mathematics' book and 'fast.ai deep learning' book.", 'duration': 27.623, 'max_score': 7494.975, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7494975.jpg'}, {'end': 7561.277, 'src': 'embed', 'start': 7533.196, 'weight': 3, 'content': [{'end': 7539.938, 'text': 'but for some reason the most common question I get is how much math or one of the most common questions I get how much math do I need to know?', 'start': 7533.196, 'duration': 6.742}, {'end': 7551.748, 'text': "They're scared of learning math because some high school teacher said they were bad at math or they didn't teach it very well and they didn't tell them that math is actually the language of nature and it's actually beautiful once you start to get into it.", 'start': 7540.378, 'duration': 11.37}, {'end': 7555.431, 'text': "they look at the greek symbols and they're like wow, i can't do all this, but you actually can.", 'start': 7551.748, 'duration': 3.683}, {'end': 7561.277, 'text': "so my approach is to start learning code first and then learn math when it's required.", 'start': 7555.431, 'duration': 5.846}], 'summary': "Many people fear math due to past experiences, but it's essential for understanding nature. learning code first, then math when needed, is a helpful approach.", 'duration': 28.081, 'max_score': 7533.196, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7533196.jpg'}, {'end': 7693.027, 'src': 'embed', 'start': 7664.194, 'weight': 4, 'content': [{'end': 7668.437, 'text': 'that says hands-on machine learning with scikit-learn and TensorFlow, Keras and TensorFlow.', 'start': 7664.194, 'duration': 4.243}, {'end': 7670.738, 'text': "To turn that into a video course, it'd be hundreds of hours.", 'start': 7668.457, 'duration': 2.281}, {'end': 7673.12, 'text': 'So learn to love reading.', 'start': 7671.479, 'duration': 1.641}, {'end': 7678.523, 'text': 'And often the latest and greatest research is published in text form, not video form.', 'start': 7674.3, 'duration': 4.223}, {'end': 7685.279, 'text': "If you're just getting started, accept their materials here to be plenty enough to keep you content for two to three years,", 'start': 7679.588, 'duration': 5.691}, {'end': 7687.584, 'text': 'the equivalent of an undergraduate degree.', 'start': 7685.279, 'duration': 2.305}, {'end': 7689.946, 'text': "Now I'm being serious with that.", 'start': 7688.905, 'duration': 1.041}, {'end': 7693.027, 'text': "You'll probably see a lot of things online like learn machine learning in six weeks.", 'start': 7689.986, 'duration': 3.041}], 'summary': 'Learning machine learning may take hundreds of hours, equivalent to an undergraduate degree.', 'duration': 28.833, 'max_score': 7664.194, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7664194.jpg'}, {'end': 7747.429, 'src': 'embed', 'start': 7701.772, 'weight': 0, 'content': [{'end': 7705.073, 'text': 'The majority of machine learning problems are actually infrastructure problems.', 'start': 7701.772, 'duration': 3.301}, {'end': 7707.735, 'text': 'So software engineering meets machine learning.', 'start': 7705.134, 'duration': 2.601}, {'end': 7711.437, 'text': "And we've covered a few things on software engineering meeting machine learning.", 'start': 7708.235, 'duration': 3.202}, {'end': 7713.518, 'text': "Remember that's called MLOps.", 'start': 7711.877, 'duration': 1.641}, {'end': 7716.299, 'text': 'So make sure to check out that blog post on MLOps.', 'start': 7713.998, 'duration': 2.301}, {'end': 7723.457, 'text': "But, Without any further ado, let's get into machine learning resources.", 'start': 7716.9, 'duration': 6.557}, {'end': 7728.682, 'text': "So if we come back, I've actually made another little cool graphic, where to start learning.", 'start': 7723.477, 'duration': 5.205}, {'end': 7731.504, 'text': 'Ooh, look at that.', 'start': 7728.702, 'duration': 2.802}, {'end': 7732.725, 'text': "That's pretty cool.", 'start': 7732.105, 'duration': 0.62}, {'end': 7738.471, 'text': "So if you're an absolute beginner, expect this little flow chart here to take three to six months.", 'start': 7733.486, 'duration': 4.985}, {'end': 7743.448, 'text': "If you're an advanced learner, you've got some familiarity with all of these.", 'start': 7739.406, 'duration': 4.042}, {'end': 7747.429, 'text': 'go through something like this expect this to take six to 12 months plus.', 'start': 7743.448, 'duration': 3.981}], 'summary': 'Machine learning problems are often infrastructure problems. mlops integrates software engineering and machine learning. beginners may take 3-6 months, advanced learners 6-12+ months.', 'duration': 45.657, 'max_score': 7701.772, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7701772.jpg'}], 'start': 6817.93, 'title': 'Machine learning fundamentals', 'summary': 'Covers the machine learning lifecycle, tools like mlflow, kubeflow, seldon, streamlit, azure machine learning, sagemaker, google cloud, and kubernetes, and emphasizes creating personalized roadmaps. it also discusses the significance of focusing on the application of machine learning tools, key mathematical concepts, and learning resources with suggested timeframes based on expertise level.', 'chapters': [{'end': 7073.705, 'start': 6817.93, 'title': 'Machine learning lifecycle', 'summary': 'Introduces the machine learning lifecycle, emphasizing tools like mlflow, kubeflow, seldon, and streamlit, as well as beginner-friendly tool streamlit, while also mentioning azure machine learning, sagemaker, google cloud, and kubernetes, and encourages creating personalized roadmaps for better understanding.', 'duration': 255.775, 'highlights': ['The chapter introduces the machine learning lifecycle', 'Tools like MLflow, Kubeflow, Seldon, and Streamlit are emphasized', 'Azure Machine Learning, SageMaker, Google Cloud, and Kubernetes are mentioned', 'Encouragement to create personalized roadmaps for better understanding']}, {'end': 7461.156, 'start': 7073.705, 'title': 'Machine learning tools and mathematics', 'summary': 'Discusses the significance of focusing on the application of machine learning tools rather than the tools themselves, and provides an overview of key mathematical concepts in machine learning, including linear algebra, calculus, probability, and optimization.', 'duration': 387.451, 'highlights': ['Machine learning is more about the application of tools rather than the tools themselves, emphasizing the significance of utilizing tools effectively.', 'The significance of mathematics in machine learning, including linear algebra, calculus, probability, and optimization, is highlighted, with an emphasis on learning these concepts as needed.', 'The chapter explains the application of linear algebra in machine learning and suggests various resources for learning this topic.']}, {'end': 7786.011, 'start': 7461.857, 'title': 'Machine learning learning resources', 'summary': 'Discusses the mathematics required for machine learning, emphasizing the importance of learning math, provides an overview of machine learning resources, and suggests timeframes for learning based on expertise level.', 'duration': 324.154, 'highlights': ['The majority of machine learning problems are infrastructure problems, where software engineering meets machine learning, and MLOps (Machine Learning Operations) is crucial, as covered in the chapter. (relevance: 5)', 'Learning machine learning resources is suggested to take 3-6 months for absolute beginners and 6-12 months plus for advanced learners, with the advice that there is no rush in learning. (relevance: 4)', "The chapter emphasizes the importance of learning math for artificial intelligence researchers and recommends resources like 'Machine Learning for Mathematics' and 'Deep Learning Book'. (relevance: 3)", 'The chapter encourages starting with code first and then learning math when required, aiming to dispel the fear of learning math and promoting the beautiful aspects of mathematics. (relevance: 2)', 'Learning materials for machine learning are provided, with the suggestion that while videos are useful, reading is essential due to the depth of content covered in books, and the prevalence of research published in text form. (relevance: 1)']}], 'duration': 968.081, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk6817930.jpg', 'highlights': ['The majority of machine learning problems are infrastructure problems, where software engineering meets machine learning, and MLOps (Machine Learning Operations) is crucial.', 'Learning machine learning resources is suggested to take 3-6 months for absolute beginners and 6-12 months plus for advanced learners, with the advice that there is no rush in learning.', "The chapter emphasizes the importance of learning math for artificial intelligence researchers and recommends resources like 'Machine Learning for Mathematics' and 'Deep Learning Book'.", 'The chapter encourages starting with code first and then learning math when required, aiming to dispel the fear of learning math and promoting the beautiful aspects of mathematics.', 'Learning materials for machine learning are provided, with the suggestion that while videos are useful, reading is essential due to the depth of content covered in books, and the prevalence of research published in text form.', 'Machine learning is more about the application of tools rather than the tools themselves, emphasizing the significance of utilizing tools effectively.', 'The chapter introduces the machine learning lifecycle', 'Tools like MLflow, Kubeflow, Seldon, and Streamlit are emphasized', 'Azure Machine Learning, SageMaker, Google Cloud, and Kubernetes are mentioned', 'Encouragement to create personalized roadmaps for better understanding', 'The chapter explains the application of linear algebra in machine learning and suggests various resources for learning this topic.']}, {'end': 8392.14, 'segs': [{'end': 7813.836, 'src': 'embed', 'start': 7786.411, 'weight': 0, 'content': [{'end': 7789.733, 'text': 'So this is a missing semester, the missing part of your computer science degree.', 'start': 7786.411, 'duration': 3.322}, {'end': 7792.074, 'text': 'This is going to teach you this little curriculum here.', 'start': 7790.013, 'duration': 2.061}, {'end': 7796.856, 'text': "We'll teach you a lot of the little parts that machine learning courses tend to miss out on.", 'start': 7792.594, 'duration': 4.262}, {'end': 7799.437, 'text': "And that's just some computer science things.", 'start': 7797.396, 'duration': 2.041}, {'end': 7809.045, 'text': "You'll probably also want to choose one cloud provider that you get familiar with, because, as you might've seen before in my Airbnb project,", 'start': 7800.097, 'duration': 8.948}, {'end': 7813.836, 'text': 'because I knew how to use Google cloud, I could get that my my code, out into the world.', 'start': 7809.045, 'duration': 4.791}], 'summary': 'This missing semester covers computer science topics and emphasizes the importance of familiarizing with a cloud provider for practical application.', 'duration': 27.425, 'max_score': 7786.411, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7786411.jpg'}, {'end': 7855.069, 'src': 'embed', 'start': 7823.029, 'weight': 2, 'content': [{'end': 7826.051, 'text': 'Khan Academy, which is great for math when needed.', 'start': 7823.029, 'duration': 3.022}, {'end': 7831.055, 'text': 'If you want to figure out or find state-of-the-art research, you probably want to visit Archive.', 'start': 7826.632, 'duration': 4.423}, {'end': 7835.78, 'text': "That's all the technical papers of computer science, physics, mathematics, and all that sort of stuff.", 'start': 7831.316, 'duration': 4.464}, {'end': 7842.463, 'text': "and if you want to version control your code, which is where you save the code that you've been writing to multiple versions.", 'start': 7836.64, 'duration': 5.823}, {'end': 7847.566, 'text': 'so if you code on day one, when it breaks on day two, you can revert back to day one.', 'start': 7842.463, 'duration': 5.103}, {'end': 7855.069, 'text': 'but if you want to add a book to this, i would highly recommend part one of the hands-on machine learning with scikit-learn,', 'start': 7847.566, 'duration': 7.503}], 'summary': "Khan academy for math, archive for research, version control for coding, and 'hands-on machine learning with scikit-learn' for book recommendation.", 'duration': 32.04, 'max_score': 7823.029, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7823029.jpg'}, {'end': 7896.249, 'src': 'embed', 'start': 7863.452, 'weight': 1, 'content': [{'end': 7866.534, 'text': 'You might go machine learning concepts, get your mind ready.', 'start': 7863.452, 'duration': 3.082}, {'end': 7872.937, 'text': "We've covered a lot of the concepts in this video and there are plenty of resources linked to the roadmap.", 'start': 7866.754, 'duration': 6.183}, {'end': 7874.818, 'text': 'So go through some of the concepts.', 'start': 7873.097, 'duration': 1.721}, {'end': 7879.96, 'text': "You're going to learn these tools, Python, within Jupyter or Google Colab.", 'start': 7875.518, 'duration': 4.442}, {'end': 7882.942, 'text': 'So Python, NumPy for numerical Python.', 'start': 7880.381, 'duration': 2.561}, {'end': 7888.044, 'text': 'So remember, machine learning is turning data into numbers and manipulating those numbers.', 'start': 7883.022, 'duration': 5.022}, {'end': 7896.249, 'text': 'A lot, a lot, and I mean a lot of data processing, of numerical processing, is based on how NumPy processes data.', 'start': 7888.685, 'duration': 7.564}], 'summary': 'Learn machine learning concepts using python and numpy for numerical processing.', 'duration': 32.797, 'max_score': 7863.452, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7863452.jpg'}, {'end': 8266.674, 'src': 'embed', 'start': 8238.34, 'weight': 3, 'content': [{'end': 8240.96, 'text': "Beginner If you're completely new, start here.", 'start': 8238.34, 'duration': 2.62}, {'end': 8241.861, 'text': 'There we go.', 'start': 8241.402, 'duration': 0.459}, {'end': 8246.9, 'text': "you're completely new to machine learning, start by learning some python code first.", 'start': 8243.357, 'duration': 3.543}, {'end': 8251.442, 'text': 'okay, then, if you want to learn python code, you can learn python in one video on youtube.', 'start': 8246.9, 'duration': 4.542}, {'end': 8252.724, 'text': 'by free code camp.', 'start': 8251.442, 'duration': 1.282}, {'end': 8255.306, 'text': 'you can do the zero to mastery python course.', 'start': 8252.724, 'duration': 2.582}, {'end': 8260.87, 'text': 'so this is full stack python taught by my business partner, andre, from the zero to mastery academy.', 'start': 8255.306, 'duration': 5.564}, {'end': 8261.851, 'text': 'look at that.', 'start': 8260.87, 'duration': 0.981}, {'end': 8263.833, 'text': 'see zero to mastery academy.', 'start': 8261.851, 'duration': 1.982}, {'end': 8265.193, 'text': 'is. this is a big disclaimer.', 'start': 8263.833, 'duration': 1.36}, {'end': 8266.674, 'text': "i'm part of this right.", 'start': 8265.193, 'duration': 1.481}], 'summary': 'Beginner machine learning: start with python, zero to mastery course on youtube.', 'duration': 28.334, 'max_score': 8238.34, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8238340.jpg'}, {'end': 8371.371, 'src': 'embed', 'start': 8323.035, 'weight': 4, 'content': [{'end': 8324.516, 'text': "We've got all these topics in here.", 'start': 8323.035, 'duration': 1.481}, {'end': 8328.156, 'text': "You'll find that on the Zero to Master Academy or on Udemy.", 'start': 8325.156, 'duration': 3}, {'end': 8335.499, 'text': 'You can also go to the Kaggle Learning Center, Faster Data Science Education.', 'start': 8329.916, 'duration': 5.583}, {'end': 8337.16, 'text': "I'm pretty sure this is all free.", 'start': 8335.839, 'duration': 1.321}, {'end': 8340.27, 'text': 'DataCamp, DataQuest.', 'start': 8338.888, 'duration': 1.382}, {'end': 8350.1, 'text': "Now if you want example projects, once you've got three to six months plus of beginner work, The next step is to go to the advanced path.", 'start': 8341.071, 'duration': 9.029}, {'end': 8356.664, 'text': "So if we go here, you want to also have done, I can't stress this enough, a milestone project.", 'start': 8350.76, 'duration': 5.904}, {'end': 8362.527, 'text': 'No matter what an instructor, including myself or including anyone from another course or whatever,', 'start': 8356.884, 'duration': 5.643}, {'end': 8368.73, 'text': "no matter how much they tell you about these things, including this video, it won't matter until you put it into practice.", 'start': 8362.527, 'duration': 6.203}, {'end': 8371.371, 'text': "I was gonna say into project, but that didn't really make sense.", 'start': 8369.05, 'duration': 2.321}], 'summary': 'Various data science resources available, including zero to master academy, udemy, kaggle learning center, datacamp, and dataquest, all offering free education. emphasizes importance of milestone projects for practical application.', 'duration': 48.336, 'max_score': 8323.035, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8323035.jpg'}], 'start': 7786.411, 'title': 'Computer science and machine learning', 'summary': 'Covers the missing semester of computer science, focusing on key components such as cloud providers, web development, machine learning concepts, python, numpy, pandas, scikit-learn, milestone projects, fast ai, tensorflow, and full stack deep learning, along with guidance on getting started with machine learning, emphasizing practical application, learning python, essential tooling, and advanced path recommendations.', 'chapters': [{'end': 8216.902, 'start': 7786.411, 'title': 'Missing semester of computer science', 'summary': 'Highlights the key components of a missing semester in computer science, including learning resources and tools such as cloud providers, web development, machine learning concepts, python, numpy, pandas, scikit-learn, milestone projects, fast ai, tensorflow, full stack deep learning, and recommended learning progression.', 'duration': 430.491, 'highlights': ['The missing semester focuses on essential components of computer science, such as machine learning concepts, cloud providers, web development, and version control, aiming to provide a comprehensive understanding of the field.', 'The recommended learning progression includes mastering Python, NumPy, Pandas, and Scikit-learn, with a focus on building milestone projects using streamlit and advancing to fast AI, TensorFlow, and full stack deep learning.', 'The chapter emphasizes the importance of hands-on experience, project-based learning, and utilizing available resources such as Khan Academy, Archive, FreeCodeCamp, and Made with ML for comprehensive skill development in computer science and machine learning.', 'The transcript provides valuable recommendations for learning resources, including Khan Academy for math, Archive for state-of-the-art research, FreeCodeCamp for web development, and Made with ML for machine learning topics and tutorials.']}, {'end': 8392.14, 'start': 8216.902, 'title': 'Getting started with machine learning', 'summary': 'Provides guidance on getting started with machine learning, including learning resources, tools, and the importance of milestone projects for beginners, while emphasizing the need for practical application. key topics include learning python, essential tooling, and advanced path recommendations.', 'duration': 175.238, 'highlights': ["Learning Python is recommended for beginners, with resources such as the 'Zero to Mastery Python Course' and 'Python like you mean it' for specialized applications like data analysis and machine learning.", 'Understanding the significance of milestone projects and practical application is emphasized as crucial for learning, illustrating the importance of hands-on experience in the learning process.', 'Recommendation to utilize learning platforms like Kaggle Learning Center, DataCamp, and DataQuest for further education and example projects after gaining three to six months of beginner-level experience.']}], 'duration': 605.729, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk7786411.jpg', 'highlights': ['The missing semester covers essential components of computer science, including machine learning concepts, cloud providers, and web development.', 'Recommended learning progression involves mastering Python, NumPy, Pandas, and Scikit-learn, focusing on milestone projects and advancing to fast AI, TensorFlow, and full stack deep learning.', 'Emphasizes hands-on experience, project-based learning, and utilizing resources like Khan Academy, Archive, FreeCodeCamp, and Made with ML for comprehensive skill development.', "Learning Python is recommended for beginners, with resources like 'Zero to Mastery Python Course' and 'Python like you mean it' for specialized applications.", 'Understanding the significance of milestone projects and practical application is emphasized as crucial for learning.', 'Recommendation to utilize learning platforms like Kaggle Learning Center, DataCamp, and DataQuest for further education and example projects after gaining three to six months of beginner-level experience.']}, {'end': 9418.728, 'segs': [{'end': 8420.343, 'src': 'embed', 'start': 8393.018, 'weight': 0, 'content': [{'end': 8396.542, 'text': 'These concepts here are the equivalent of your mother and father telling you that the stove is hot.', 'start': 8393.018, 'duration': 3.524}, {'end': 8404.289, 'text': "The project is you touching the stove and figuring out, okay, you've told me all these things, now it's my turn to touch the stove.", 'start': 8397.002, 'duration': 7.287}, {'end': 8411.457, 'text': "So once you've done some beginner stuff, come into the advanced, have a look at some end-to-end projects, what they look like.", 'start': 8405.871, 'duration': 5.586}, {'end': 8414.339, 'text': 'This is probably before you got into the advanced stuff here.', 'start': 8411.497, 'duration': 2.842}, {'end': 8420.343, 'text': "is probably what you'd want to be working on is some projects like these.", 'start': 8415.259, 'duration': 5.084}], 'summary': 'Encouraging exploration through hands-on projects, progressing from beginner to advanced levels.', 'duration': 27.325, 'max_score': 8393.018, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8393018.jpg'}, {'end': 8582.873, 'src': 'embed', 'start': 8553.213, 'weight': 1, 'content': [{'end': 8554.594, 'text': 'Cannot recommend this book enough.', 'start': 8553.213, 'duration': 1.381}, {'end': 8557.156, 'text': 'Hands-on machine learning with scikit-learn and TensorFlow.', 'start': 8554.874, 'duration': 2.282}, {'end': 8559.258, 'text': "I've literally got that sitting right next to me.", 'start': 8557.477, 'duration': 1.781}, {'end': 8559.999, 'text': 'See, there we go.', 'start': 8559.318, 'duration': 0.681}, {'end': 8565.848, 'text': 'You last purchased this item on 11th of April, 2020.', 'start': 8560.419, 'duration': 5.429}, {'end': 8572.11, 'text': 'Deep Learning for Coders by the fast AI teachers, Jeremy Howard and Silvian Gugga.', 'start': 8565.848, 'duration': 6.262}, {'end': 8574.131, 'text': "That's coming out in July, 2020.", 'start': 8572.63, 'duration': 1.501}, {'end': 8575.871, 'text': 'So this is actually a preview of this book.', 'start': 8574.131, 'duration': 1.74}, {'end': 8576.911, 'text': 'Look at that.', 'start': 8576.571, 'duration': 0.34}, {'end': 8579.612, 'text': "You come to this video and you get all these things that aren't even out yet.", 'start': 8576.991, 'duration': 2.621}, {'end': 8580.993, 'text': 'Come on, Daniel.', 'start': 8580.352, 'duration': 0.641}, {'end': 8582.873, 'text': 'Building Machine Learning Pipelines.', 'start': 8581.493, 'duration': 1.38}], 'summary': 'Enthusiastic recommendation for hands-on machine learning books and upcoming deep learning book by fast ai teachers.', 'duration': 29.66, 'max_score': 8553.213, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8553213.jpg'}, {'end': 8684.572, 'src': 'embed', 'start': 8657.751, 'weight': 2, 'content': [{'end': 8665.175, 'text': "So if you're watching this video and thinking that I'm some sort of expert on all of this stuff, please don't get too far ahead of yourself.", 'start': 8657.751, 'duration': 7.424}, {'end': 8666.956, 'text': "I'm still learning all of this as well.", 'start': 8665.375, 'duration': 1.581}, {'end': 8671.88, 'text': 'If you wanna learn a cloud service, a Cloud Guru is probably one of the best places to go.', 'start': 8667.756, 'duration': 4.124}, {'end': 8676.204, 'text': "I've done a few courses on there, especially getting certified with Google Cloud.", 'start': 8671.96, 'duration': 4.244}, {'end': 8677.666, 'text': 'Highly recommend that.', 'start': 8676.765, 'duration': 0.901}, {'end': 8684.572, 'text': 'And then Google Cloud, if you wanna learn that, AWS, Microsoft Azure.', 'start': 8679.447, 'duration': 5.125}], 'summary': 'Cloud guru is recommended for learning cloud services, especially for google cloud certification.', 'duration': 26.821, 'max_score': 8657.751, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8657751.jpg'}, {'end': 8862.7, 'src': 'embed', 'start': 8837.061, 'weight': 4, 'content': [{'end': 8842.705, 'text': "And what you'll be surprised to find is that a lot of people probably have the same questions that you do.", 'start': 8837.061, 'duration': 5.644}, {'end': 8848.189, 'text': 'So if you want to create a blog, try fast pages or GitHub pages or medium.', 'start': 8843.085, 'duration': 5.104}, {'end': 8855.515, 'text': "There's a whole bunch of different reasons that you can try, but what writing does is it shows you how poor your thinking is.", 'start': 8848.81, 'duration': 6.705}, {'end': 8860.299, 'text': 'So when you think you understand something, try write about it and teach someone else about it.', 'start': 8856.035, 'duration': 4.264}, {'end': 8862.7, 'text': "That's when you'll really start to understand it.", 'start': 8861.039, 'duration': 1.661}], 'summary': 'Writing helps improve understanding. use fast pages, github, or medium for blogging.', 'duration': 25.639, 'max_score': 8837.061, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8837061.jpg'}, {'end': 8912.75, 'src': 'embed', 'start': 8876.249, 'weight': 3, 'content': [{'end': 8879.792, 'text': 'Made with ML, so this is a community-driven resource for your projects.', 'start': 8876.249, 'duration': 3.543}, {'end': 8881.114, 'text': "We've actually just been through that.", 'start': 8879.892, 'duration': 1.222}, {'end': 8884.597, 'text': 'But if you do make a machine learning project, you should definitely post it there.', 'start': 8881.354, 'duration': 3.243}, {'end': 8891.143, 'text': "whether it's something as simple as a blog post of 10 things I learned in my first machine learning course should put it there,", 'start': 8884.597, 'duration': 6.546}, {'end': 8893.485, 'text': "or whether it's something phenomenal.", 'start': 8891.143, 'duration': 2.342}, {'end': 8897.529, 'text': "I'm not saying that 10 things I learned during my first machine learning course isn't phenomenal,", 'start': 8893.845, 'duration': 3.684}, {'end': 8900.672, 'text': "or whether it's something crazy like I turned my car into a self-driving car.", 'start': 8897.529, 'duration': 3.143}, {'end': 8903.655, 'text': "You should put whatever you're doing, put it there.", 'start': 8901.411, 'duration': 2.244}, {'end': 8905.077, 'text': 'Everyone has to start somewhere.', 'start': 8903.855, 'duration': 1.222}, {'end': 8906.279, 'text': 'I cannot stress this enough.', 'start': 8905.117, 'duration': 1.162}, {'end': 8908.563, 'text': 'So we come back to the keynote.', 'start': 8907.201, 'duration': 1.362}, {'end': 8912.75, 'text': 'Where are we up to? Oh, example curriculums.', 'start': 8909.825, 'duration': 2.925}], 'summary': 'Community-driven resource for ml projects, emphasizes sharing and starting somewhere.', 'duration': 36.501, 'max_score': 8876.249, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8876249.jpg'}, {'end': 9049.799, 'src': 'embed', 'start': 9021.809, 'weight': 5, 'content': [{'end': 9024.27, 'text': 'SOTA Bench, Papers with Code, Made with ML.', 'start': 9021.809, 'duration': 2.461}, {'end': 9031.872, 'text': "So SOTA Bench is where you're gonna find all of the state of the art machine learning models benchmarked on a number of different data sets.", 'start': 9024.65, 'duration': 7.222}, {'end': 9038.915, 'text': "Papers with Code is where you'll get the latest machine learning research with code attached, usually.", 'start': 9032.512, 'duration': 6.403}, {'end': 9041.776, 'text': "So, although it's called Papers with Code,", 'start': 9039.575, 'duration': 2.201}, {'end': 9047.258, 'text': 'I think one of the criteria actually for something to come up on Papers with Code is that it has to be a research paper,', 'start': 9041.776, 'duration': 5.482}, {'end': 9049.799, 'text': 'so like state-of-the-art machine learning, but with code.', 'start': 9047.258, 'duration': 2.541}], 'summary': 'Sota bench and papers with code host state-of-the-art ml models and research papers with code.', 'duration': 27.99, 'max_score': 9021.809, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk9021809.jpg'}], 'start': 8393.018, 'title': 'Machine learning and resources', 'summary': 'Covers the importance of hands-on projects, recommended learning paths, and additional resources for mastering machine learning and related disciplines. it emphasizes practical learning, recommends online resources, and encourages creating and sharing learning experiences. it also includes various resources for machine learning, data sources, learning courses, and emphasizes the importance of exploring and gradually deepening understanding in machine learning.', 'chapters': [{'end': 8657.251, 'start': 8393.018, 'title': 'Machine learning journey', 'summary': 'Emphasizes the importance of hands-on projects, advises on recommended learning paths, and suggests additional resources for mastering machine learning and related disciplines.', 'duration': 264.233, 'highlights': ['The chapter stresses the significance of hands-on projects and recommends transitioning from beginner to advanced level through end-to-end projects, marking a journey of three to six months.', 'It advises concurrent enrollment in deeplearning.ai and fast.ai curriculums and emphasizes the importance of not getting fixated on the latest tools, but rather building projects that deliver value.', 'Encourages proficiency in at least one cloud service and underscores the significance of general software engineering practices, mathematics, and Python for machine learning, along with recommended resources for further learning.', "The chapter provides insights into recommended books, including 'Automate the Boring Stuff with Python' and 'Hands-On Machine Learning with scikit-learn and TensorFlow', and previews upcoming books, such as 'Deep Learning for Coders' and 'Building Machine Learning Pipelines'.", "It also advocates for understanding the mathematics behind machine learning, recommends 'Mathematics for Machine Learning' as a vital read, and suggests additional resources for learning matrix calculus and neural network principles."]}, {'end': 9021.369, 'start': 8657.751, 'title': 'Machine learning tips and resources', 'summary': 'Emphasizes the importance of practical learning, recommending online resources like a cloud guru for cloud services, and encourages creating and sharing learning experiences through blog posts and community-driven platforms like made with ml.', 'duration': 363.618, 'highlights': ['A Cloud Guru is recommended for learning cloud services, especially for getting certified with Google Cloud.', 'Emphasizes the importance of creating and sharing learning experiences through blog posts and community-driven platforms like Made with ML.', 'Suggests exploring useful tools like Fastpages, GitHub pages, and Medium for creating a blog and emphasizes the importance of writing to enhance understanding.']}, {'end': 9418.728, 'start': 9021.809, 'title': 'Machine learning roadmap 2020', 'summary': 'Covers various resources for machine learning including sota bench, papers with code, and made with ml, along with data sources like google dataset search and kaggle datasets. it also includes recommendations for learning courses and emphasizes the importance of exploring and gradually deepening understanding in machine learning.', 'duration': 396.919, 'highlights': ['Covered various resources for machine learning including SOTA Bench, Papers with Code, and Made with ML.', 'Included recommendations for learning courses and emphasized the importance of exploring and gradually deepening understanding in machine learning.', 'Provided data sources such as Google Dataset Search and Kaggle Datasets.']}], 'duration': 1025.71, 'thumbnail': 'https://coursnap.oss-ap-southeast-1.aliyuncs.com/video-capture/pHiMN_gy9mk/pics/pHiMN_gy9mk8393018.jpg', 'highlights': ['The chapter stresses the significance of hands-on projects and recommends transitioning from beginner to advanced level through end-to-end projects, marking a journey of three to six months.', "The chapter provides insights into recommended books, including 'Automate the Boring Stuff with Python' and 'Hands-On Machine Learning with scikit-learn and TensorFlow', and previews upcoming books, such as 'Deep Learning for Coders' and 'Building Machine Learning Pipelines'.", 'A Cloud Guru is recommended for learning cloud services, especially for getting certified with Google Cloud.', 'Emphasizes the importance of creating and sharing learning experiences through blog posts and community-driven platforms like Made with ML.', 'Suggests exploring useful tools like Fastpages, GitHub pages, and Medium for creating a blog and emphasizes the importance of writing to enhance understanding.', 'Covered various resources for machine learning including SOTA Bench, Papers with Code, and Made with ML.', 'Included recommendations for learning courses and emphasized the importance of exploring and gradually deepening understanding in machine learning.']}], 'highlights': ['Neural networks represent a fundamental shift in software writing, referred to as software 2.0.', 'Machine learning involves turning data into numbers and finding patterns.', 'Machine learning roadmap for 2020 emphasizes an interactive living mind map and compass.', 'Transfer learning saves significant training time, such as 70,000 hours for a self-driving car model.', 'Evaluation metrics like confusion matrix provide a method to evaluate ML model performance.', 'R squared, mean squared error, and mean absolute error are commonly used in regression problems.', 'Feature engineering involves encoding domain knowledge into data for more meaningful representations.', 'The curse of dimensionality refers to phenomena in high dimensional spaces.', "Tools and processes for deploying ML models, including TensorFlow Serving, PyTorch Serving, Google's AI platform, SageMaker, and MLOps, are explored.", 'Cloud compute services like Google Cloud, AWS, and Microsoft Azure are essential for larger compute needs.', 'Learning machine learning resources is suggested to take 3-6 months for absolute beginners and 6-12 months plus for advanced learners.', 'The missing semester covers essential components of computer science, including machine learning concepts, cloud providers, and web development.', 'The chapter stresses the significance of hands-on projects and recommends transitioning from beginner to advanced level through end-to-end projects.', 'A Cloud Guru is recommended for learning cloud services, especially for getting certified with Google Cloud.', 'Emphasizes the importance of creating and sharing learning experiences through blog posts and community-driven platforms like Made with ML.']}