Chevron Left
Volver a Practical Reinforcement Learning

Practical Reinforcement Learning, National Research University Higher School of Economics

4.3
115 calificaciones
33 revisiones

Acerca de este Curso

Welcome to the Reinforcement Learning course. Here you will find out about: - foundations of RL methods: value/policy iteration, q-learning, policy gradient, etc. --- with math & batteries included - using deep neural networks for RL tasks --- also known as "the hype train" - state of the art RL algorithms --- and how to apply duct tape to them for practical problems. - and, of course, teaching your neural network to play games --- because that's what everyone thinks RL is about. We'll also use it for seq2seq and contextual bandits. Jump in. It's gonna be fun!...

Principales revisiones

por TC

May 17, 2018

Great course. Best course so far on reinforcement learning.

Filtrar por:

33 revisiones

por Pedro Luís Alves Veloso

Nov 27, 2018

Pros:

-It is a pioneer RL course in Coursera.

-Great exercise templates with interesting applications of RL algorithms.

-There are always references to good papers and new developments in RL.

-Good sense of humor in the lecture and templates.

-The discussion forum addresses the the bugs of the course.

-The course is challenging in the right level.

Cons:

-The lectures are not in that level yet ... they do not explain the important parts in detail.

-The lecturers should improve their public speaking and storytelling skills.

-The course subverts the sequence of the RL topics (cross-entropy is the first method and the multi-armed bandits setting is in the last week). This could be good, but ended up being confusing.

-The quizzes and exercises still contain many bugs.

Overall:

This is a good course, but it has the potential to be much better. If you want to challenge yourself and solve really interesting problems, take this course. You will probably have to watch David Silver's lectures on YouTube and read some parts of Sutton and Barto's book to understand the concepts. However, if you feel frustrated dealing with bugs in the exercises or answering quizzes that are confusing, do not take this course.

por Roman Puchkovskii

Nov 05, 2018

The course is really in 'beta' state. Be prepared to struggle against not only the practical assignments themselves, but also against their bugs and assignment grading infrastructure problems.

But the course content itself is very useful and worth the trouble. Also, most of the bugs and problems are already solved by the community, you just need to visit the Discussion forums to find the solutions.

por Guy Koren

Nov 04, 2018

great content !

administration could benefit from some improvements (some exercises required "hacking" but the course forum were helpful)

Also, would be great if the slides can be shared.

this is the 2nd course I take from HSE. very happy with the content and the level. exercises are excellent !

I will happily continue to the next course in this specialization :)

por maciej.osinski

Nov 02, 2018

Brilliant content but quite some bugs in assignments

por Jay Glascoe

Oct 30, 2018

Course 4 of Advanced Machine Learning, Practical Reinforcement Learning, is harder than Course 1, Introduction to Deep Learning. (I jumped to Course 4 after Course 1). That is saying quite a lot because I would describe Course 1 as "fiendishly difficult".

There's a few reasons for why 4 is harder than 1.

One big reason is, the course is still "in beta". Not everything, and maybe not anything, works as a straightforward Coursera Notebook. My workaround was to download the courses as IPYNB files, and then upload them to Google Colab. I'm glad for the experience as I'm now very familiar with Google Colab and how to navigate a Coursera notebook environment to get at the grader.py and submit.py files needed.

If you are not at least somewhat skilled as a programmer, you may want to avoid this course until it is out of beta.

Second reason is the Quizzes. These quizzes, most of them, are difficult. I myself never resorted to "try every possible permutation" to pass a quiz, but I did have to retake quizzes, re-watch videos, Ctrl-F find words in the video Interactive Transcript area, and read the Forums for help. Get ready to have some "fun" (and by "fun" I mean the opposite of "fun") taking these quizzes.

Third reason is, Alexander Panin can occasionally be difficult to understand in English (that's as gently as I can put it). But this, too, I'm glad for the experience. The neural networks in my brain for translating "thick Russian accent" to "colloquial English" have improved greatly. But everyone should take it easy on Alexander, because...

This course of his is awesome! I dreaded the Videos. I hated the Quizzes. And the assignments? Until I had finished an assignment 100%, it was the bane of my existence. But when you solve the assignment? Exhilerating. The assignments are a treasure trove of HOW-TOs on different RL techniques. Have you got an RL problem you want to solve? Chances are at least one of these notebooks will either flat out give you the solution, or else at least point the way forward.

por Sergey

Oct 13, 2018

Доведите ноутбуки и grader до ума, не позорьтесь пожалуйста!

por Kota Mori

Oct 04, 2018

The class is very immature as of September 2018. A good reason for taking this course is because it is one of few online courses where you can play with actual programming exercises of various reinforcement learning techniques, from dynamic programming to deep Q networks and actor critiques. Examples are mostly for environments of Open AI gym. You can also see examples where you use libraries such as tensorflow and pytorch used in the framework. However, the codes, including submission and grading system, have numerous bugs, which forces you to do extra debugging works unrelated to the course topics. Fortunately some early takers of the class left helpful comments on the forum, with which you can solve the most of issues if you read them carefully.

Quality of presentation is not as good as other courses I found in the Coursera. Most of the time, the lecturer seems to be just reading the scripts. To make it worse, the scripts are not written in spoken language.

por SagarSrinivas

Sep 29, 2018

Awesome. Worth it!

por RAM HEGDE

Sep 16, 2018

This is one of the best courses I have taken on Coursera, The course is very rich in content and methodically developed. Both lectures have done a wonderful job in delivering lecture with full energy and make difficult concepts graspable. Both programming assignments and quizzes are well thought out. My sincere appreciation and thanks.

por Abhilash VJ

Sep 14, 2018

This is a great introduction to reinforcement learning.I faced some problems in submitting assignments but the course content and the extra materials are really good.I think it is to RL what Andrew Ng's Course is to ML . You will implement algorithms like cross entropy method to DQNs and A3C .Assignments uses Open AI Gym so you can get some good practical results .Overall I loved this course and would like to recommend to any one who is getting started in Reinforcement Learning.

Thank you.