1 00:00:00,004 --> 00:00:03,840 Welcome to the final video of this Machine Learning class. 2 00:00:03,840 --> 00:00:06,473 We've been through a lot of different videos together. 3 00:00:06,473 --> 00:00:08,774 In this video I would like to just quickly 4 00:00:08,774 --> 00:00:11,003 summarize the main topics of this course 5 00:00:11,003 --> 00:00:13,089 and then say a few words at the end and that 6 00:00:13,089 --> 00:00:14,729 will wrap up the class. 7 00:00:16,390 --> 00:00:18,020 So what have we done? 8 00:00:18,020 --> 00:00:21,957 In this class we spent a lot of time talking about supervised learning algorithms 9 00:00:21,957 --> 00:00:25,436 like linear regression, logistic regression, neural networks, SVMs. 10 00:00:25,436 --> 00:00:29,435 for problems where you have labelled data and labelled examples 11 00:00:29,435 --> 00:00:31,300 like x(i), y(i) 12 00:00:31,300 --> 00:00:35,715 And we also spent quite a lot of time talking about unsupervised learning 13 00:00:35,715 --> 00:00:37,344 like K-means clustering, 14 00:00:37,344 --> 00:00:40,316 Principal Components Analysis for dimensionality reduction 15 00:00:40,316 --> 00:00:43,847 and Anomaly Detection algorithms for when you have only 16 00:00:43,847 --> 00:00:46,363 unlabelled data x(i) 17 00:00:46,363 --> 00:00:49,378 Although Anomaly Detection can also use some labelled data 18 00:00:49,378 --> 00:00:51,189 to evaluate the algorithm. 19 00:00:51,451 --> 00:00:54,725 We also spent some time talking about special applications 20 00:00:54,725 --> 00:00:56,407 or special topics like Recommender Systems 21 00:00:56,407 --> 00:00:58,895 and large scale machine learning systems 22 00:00:58,895 --> 00:01:01,477 including parallelized and rapid-use systems 23 00:01:01,477 --> 00:01:03,925 as well as some special applications like 24 00:01:03,925 --> 00:01:07,609 sliding windows object classification for computer vision. 25 00:01:07,609 --> 00:01:11,549 And finally we also spent a lot of time talking about different aspects 26 00:01:11,549 --> 00:01:15,198 of, sort of, advice on building a machine learning system. 27 00:01:15,198 --> 00:01:17,264 And this involved both trying to understand 28 00:01:17,264 --> 00:01:19,233 what is it that makes a machine learning algorithm 29 00:01:19,233 --> 00:01:20,561 work or not work. 30 00:01:20,561 --> 00:01:22,012 So we talked about things like bias and variance, 31 00:01:22,012 --> 00:01:25,479 and how regularization can help with some variance problems. 32 00:01:25,479 --> 00:01:28,445 And we also spent a little bit of time talking about 33 00:01:28,445 --> 00:01:32,313 this question of how to decide what to work on next. 34 00:01:32,313 --> 00:01:35,019 So, how to prioritize how you spend your time 35 00:01:35,019 --> 00:01:37,513 when you're developing a machine learning system. 36 00:01:38,021 --> 00:01:41,044 So we talked about evaluation of learning algorithms, 37 00:01:41,044 --> 00:01:44,221 evaluation metrics like precision recall, F1 score 38 00:01:44,221 --> 00:01:47,072 as well as practical aspects of evaluation 39 00:01:47,072 --> 00:01:49,898 like the training, cross-validation and test sets. 40 00:01:49,898 --> 00:01:52,319 And we also spent a lot of time talking about 41 00:01:52,319 --> 00:01:55,741 debugging learning algorithms and making sure 42 00:01:55,741 --> 00:01:57,212 the learning algorithm is working. 43 00:01:57,212 --> 00:01:59,075 So we talked about diagnostics 44 00:01:59,075 --> 00:02:01,999 like learning curves and also talked about things like 45 00:02:01,999 --> 00:02:04,394 error analysis and ceiling analysis. 46 00:02:04,394 --> 00:02:08,187 And so all of these were different tools for helping you to decide 47 00:02:08,187 --> 00:02:10,349 what to do next and how to spend your valuable 48 00:02:10,349 --> 00:02:12,585 time when you're developing a machine learning system. 49 00:02:12,585 --> 00:02:17,665 And in addition to having the tools of machine learning at your disposal 50 00:02:17,665 --> 00:02:20,228 so knowing the tools of machine learning like 51 00:02:20,228 --> 00:02:22,127 supervised learning and unsupervised learning and so on, 52 00:02:22,127 --> 00:02:26,015 I hope that you now not only have the tools, 53 00:02:26,015 --> 00:02:29,457 but that you know how to apply these tools really well 54 00:02:29,457 --> 00:02:32,658 to build powerful machine learning systems. 55 00:02:33,658 --> 00:02:35,556 So, that's it. 56 00:02:35,556 --> 00:02:37,645 Those were the topics of this class 57 00:02:37,645 --> 00:02:39,614 and if you worked all the way through this course 58 00:02:39,614 --> 00:02:41,308 you should now consider yourself 59 00:02:41,308 --> 00:02:43,511 an expert in machine learning. 60 00:02:43,511 --> 00:02:46,879 As you know, machine learning is a technology 61 00:02:46,879 --> 00:02:49,916 that's having huge impact on science, technology and industry. 62 00:02:49,916 --> 00:02:53,360 And you're now well qualified to use these tools 63 00:02:53,360 --> 00:02:55,351 of machine learning to great effect. 64 00:02:55,351 --> 00:02:57,910 I hope that many of you in this class 65 00:02:57,910 --> 00:02:59,765 will find ways to use machine learning 66 00:02:59,765 --> 00:03:02,324 to build cool systems and cool applications 67 00:03:02,324 --> 00:03:03,946 and cool products. 68 00:03:03,946 --> 00:03:06,084 And I hope that you find ways 69 00:03:06,084 --> 00:03:07,930 to use machine learning not only 70 00:03:07,930 --> 00:03:09,762 to make your life better but maybe someday 71 00:03:09,762 --> 00:03:14,749 to use it to make many other people's life better as well. 72 00:03:14,780 --> 00:03:19,699 I also wanted to let you know that this class has been great fun for me to teach. 73 00:03:19,699 --> 00:03:21,788 So, thank you for that. 74 00:03:21,788 --> 00:03:23,807 And before wrapping up, 75 00:03:23,807 --> 00:03:25,282 there's just one last thing I wanted to say. 76 00:03:25,282 --> 00:03:28,956 Which is that: It was maybe not so long ago, 77 00:03:28,956 --> 00:03:31,306 that I was a student myself. 78 00:03:31,306 --> 00:03:34,711 And even today, you know, I still try to take different courses 79 00:03:34,711 --> 00:03:36,902 when I have time to try to learn new things. 80 00:03:36,902 --> 00:03:39,989 And so I know how time-consuming it is 81 00:03:39,989 --> 00:03:42,273 to learn this stuff. 82 00:03:42,273 --> 00:03:44,663 I know that you're probably a busy person 83 00:03:44,663 --> 00:03:47,302 with many, many other things going on in your life. 84 00:03:47,302 --> 00:03:49,838 And so the fact that you still found 85 00:03:49,838 --> 00:03:52,431 the time or took the time to watch these videos 86 00:03:52,431 --> 00:03:55,799 and, you know, many of these videos just went on 87 00:03:55,799 --> 00:03:57,598 for hours, right? 88 00:03:57,598 --> 00:04:00,068 And the fact many of you took the time 89 00:04:00,068 --> 00:04:01,826 to go through the review questions 90 00:04:01,826 --> 00:04:03,731 and that many of you took the time 91 00:04:03,731 --> 00:04:06,250 to work through the programming exercises. 92 00:04:06,250 --> 00:04:09,483 And these were long and complicate programming exercises. 93 00:04:09,483 --> 00:04:12,840 I wanted to say thank you for that. 94 00:04:12,840 --> 00:04:17,920 And I know that many of you have worked hard on this class 95 00:04:17,920 --> 00:04:21,880 and that many of you have put a lot of time into this class, 96 00:04:21,880 --> 00:04:25,396 that many of you have put a lot of yourselves into this class. 97 00:04:25,396 --> 00:04:29,292 So I hope that you also got a lot of out this class. 98 00:04:29,292 --> 00:04:31,347 And I wanted to say: 99 00:04:31,347 --> 00:04:36,423 Thank you very much for having been a student in this class.