Stanford 19th Lesson: summary (Conclusion)

19.1 Summary and acknowledgements

Welcome to the last video on machine learning. We have been studying together for a long time. In the final video, I want to take a quick look at the main content of this course, and then briefly say a few words to say.

As the end time of this course, what have we learned? In this course, we spent a great deal of time introducing some supervised learning algorithms such as linear regression, logistic regression, neural network, support vector machines, and so on, which have tagged data and samples such as x (i), Y (i).

Then we spent a lot of time introducing unsupervised learning. For example, K-means clustering, principal component analysis for dimensionality reduction,

And the anomaly detection algorithm when you have only a series of untagged data x (i). Of course, sometimes tagged data can also be used for the evaluation of anomaly detection algorithms. In addition, we also take the time to discuss

Special applications or special topics, such as referral systems. As well as large-scale machine learning systems, including parallel systems and mapping simplification methods, there are some other special applications. For example, a sliding window classification algorithm for computer vision technology.

Finally, we mention a lot of practical suggestions for building machine learning systems. This includes how to understand why a machine learning algorithm works, so we talked about deviations and variances, and we talked about regularization of the problem of variance, and we discussed how to decide what to do next, that is, when you are developing a machine learning system, What work is the next priority should be given. Therefore, we discuss the evaluation method of learning algorithm. The evaluation matrix is introduced, such as: precision ratio, recall rate and F1 score, and the evaluation Learning algorithm is more practical.

Training sets, cross-validation sets, and test sets. We also introduced the debugging of learning algorithms, and how to ensure the normal operation of learning algorithms, so we introduced some diagnostic methods, such as learning curve, but also discussed the error analysis, upper limit analysis and so on.

All of these tools can effectively guide you in deciding what to do next, allowing you to spend your precious time on the edge. Now you have mastered a lot of machine learning tools, including supervised learning algorithms and unsupervised learning algorithms, and so on.

But beyond that, I would prefer that you not only know these tools now, but more importantly how to use them effectively to build a powerful machine learning system. So, this is the whole subject of this course. If you follow our course all the way up to now, you should already feel that you have become an expert in machine learning?

As we all know, machine learning is an important discipline that has a profound impact on technology and industry, and now you have the ability to apply these machine learning tools to create great achievements. I hope that many of you will be able to apply the machine learning tools you have learned in the appropriate fields to build the perfect machine learning system and develop unmatched products and applications. And I also hope that you learn through the application of machine, not only to change their lives, one day, but also to let more people live a better life!

I also want to tell you that teaching this course is a pleasure for me. So, thank you all! Finally, before I finish, I would like to say a little more: that is, maybe I was a student not long ago, even now, I try to squeeze out the time to listen to some lessons and learn something new. So, I know that it takes a while to finish this course, and I knew that maybe you are a very busy person, there are many things to deal with in life. That's why you still squeeze in the time to watch these course videos. I know that many videos take hours and you still spend a lot of time doing these review questions. Many of you are also willing to take the time to study programming exercises that are long and complex. I express my heartfelt thanks to you! I know many of you have worked very hard in this class and many people have spent a lot of time in this class, and many people have contributed a lot of their energy to this course. Therefore, I sincerely hope that you can gain from this course!

Finally, I want to say! Thank you again for taking this course!

Andew Ng

Stanford 19th Lesson: summary (Conclusion)