Hacker News Comments on
What is Reinforcement Learning? | Udacity Free Courses
Udacity
·
4
HN comments
- This course is unranked · view top recommended courses
Hacker News Stories and Comments
All the comments and stories posted to Hacker News that reference this url.As a self contained, foundational course, Georgia Tech's OMSCS offering [1] is solid. Charles Isbell and Michael Littman are great at building intuition into equations.[1] https://www.udacity.com/course/reinforcement-learning--ud600
⬐ joshuamortonIsbell's course in person was great. And if the exams for the online version are anything like the in person ones, it really does test your understanding of foundational concepts.⬐ richfnelsonYup, just took the online RL class and the average grade for the final exam was 45 out of 100, high score of 76. The format was true/false with a short explanation for your answer. I never thought I'd be proud about getting a 53 on a true/false exam, but it was an extremely challenging and rewarding class.
I've taken 10 GaTech OMSCS courses via Udacity. Two of my favorites (so far): Intro to Computer Vision: https://www.udacity.com/course/introduction-to-computer-visi... Reinforcement Learning: https://www.udacity.com/course/reinforcement-learning--ud600I found the Isbell+Littman combo to work so well that I also took the ML course. I know some people complain about their humor but it was perfect for me. I could listen to those two explain just about anything. I still LOL when I think about Littman saying to Isbell something like "are you trying to teach us something by making this lecture infinitely long?" Who knew RL could be funny?
⬐ canistrI can also attest to CV being a great OMSCS class. To me they are by far the best lectures in the program in terms of breadth and depth. They're the most similar to in-class instruction of a university unlike every other class in the program.As for Isbell+Littman... while their lectures may be more jovial, I wouldn't say they were as effective in learning. I hated the videos tbh. For ML, I found my learning consisted of watching other available videos on the web. Including institutions like CMU, UW, Stanford, and YouTube.
⬐ popekoI'm currently taking Intro to CV, and can attest it is indeed quite superb - the professor has a fantastic way of explaining theory, and goes over a really large number of topics. His antics with the videographer are also quite hilarious⬐ allanbreyesSecond, this! I was a big fan of Isbell+Littman. There was a brief conversation on Twitter about a third class[0]. Really hoping it happens!This was also probably my favorite OMSCS class. The projects were particularly enjoyable... especially the OpenAI Gym lunar lander[1]. Kinda bummed that OpenAI chose to shut down the online submission platform.
[0]: https://twitter.com/isbellHFh/status/893216499439210497
The last two lectures in the The Georgia Tech Machine Learning course on Udacity cover some basics of Game Theory. Just skip ahead, the game theory part is mostly self-contained. https://www.udacity.com/course/machine-learning--ud262 The Reinforcement Learning course includes some of the same (exactly the same) game theory content and then adds an additional lecture on further topics in game theory https://www.udacity.com/course/reinforcement-learning--ud600 Again, skip to the last couple lectures.
+1 to Sutton & Barto. There is a second edition of their book available; see a link at https://webdocs.cs.ualberta.ca/~sutton/book/the-book.html.https://www.udacity.com/course/reinforcement-learning--ud600 is also good; it is not super-modern, but it is easily accessible, and it covers a large part of classical RL - this helps if you want to read recent RL papers because terminology and ideas become more familiar.
I'm not sure Andrey Karpaty's blog post (linked in the article) is a good intro to RL - it starts with Policy Gradient which is on a complex side of RL techniques spectrum. From other sources I've heard Policy Gradient is harder to get working on an arbitrary problem than e.g. Q-Learning, but don't quote me on that :)