This is one of the Best Course available on Reinforcement Learning. I have gone through various study material but the depth and practical knowledge given in the course is awesome.
At times it felt like a bit more video material would be helpful to better understand the subject/gain deeper understanding.\n\nAnd fixing some of the notebooks would be helpful.
by Krishna H•
by Keanu T•
by Nguyen, Q H (•
I learned a lot from this course despite the very strong accent (bro please speak slowly). Most of the time I had to watch David Silver's lectures to gain a better understanding of the subject. RL is a very challenging area from both theoretical and applied perspective (at least for me it is clearly not easy), so don't expect it be a piece of cake like many of Andrew's courses on coursera. I have taken courses in probability theory, computational inference, stochastic processes and algorithms analysis, which are extremely essential to fully understand the materials in this course and RL courses in general. Assignments are challenging and very interesting but most of the heavy lifting were taken cared of. My definition of learning is that I should not expect the lecturer to take care of everything for me, they're there to give me direction and the rest is my job to find the answer that matters most to me. The teaching in this course is no where compared to Andrew's level of teaching but it will give you a very clear roadmap to further deepen your curiosity in RL field. best of luck.
by Nahas P•
Good course that covers a lot of Reinforcement Learning concepts and methods in a format that is simple and non-intimidating. It touches upon the basics of RL and Q Learning, then follows it up with explanation of popular methods like REINFORCE and MCTS.
The assignments using partially completed Jupyter notebooks reinforce the theoretical knowledge while ensuring the students are not encumbered by environment setup or non-core issues.
Both the course content and assignments progress linearly, so it was easy to follow.
A important suggestion for improvement would be to tweak the presentation style to reduce monotony. Improving animation in the slides or highlighting sections being discussed might help here.
Considering the word 'Practical' in the title, a couple of real-world applications of RL should have been part of the course as a coding examples or assignments.
While the simplicity of the assignments help in easy understanding of the topic, completion of the assignments do not impart the required confidence for handling more complex problems.
by Cristian M B B•
Is a good course on reinforcement learning. Assignments are very well structured for a virtual platform, and the explanation in video lectures are enough to solve the notebooks (probably it takes more time than the estimated but it's acceptable). Alexander's Panin Pronunciation is a little bit difficult to understand even if you decrease the video's speed. Subtitles don't help very much because it seems autogenerated and are a totally disaster in some parts. Alexander Panin is cool, he shows that he not only reads but domain the topics. However he should consider writting the subtitles of his videos himself.
Excellent course, congrats!
by Guining O•
I'd say Pareto principle really does hold here. Most of the learning part comes from the programming assignments and I'd always recommend David Silver's videos to actually understand the material. Go for this one only if you have the prerequisites ready and are willing to take up hard assignments. I started this course long ago but it was way too hard for me back then. So, I took up David Silver's course and then returned to this one for programming exercises and I must say I really liked it. Got used to Alexander Panin's accent and 1.3x his speed too. The other lecturer is better at 1.5x
by Thomas D•
The course is dense and is accompanied by quality support (references to other courses, articles,...). It is punctuated with quizzes (which are unfortunately often quite ambiguous) and exercises on jupyter (which are well guided). This course seems to me, alone, insufficient and it is necessary to go to consult some references proposed to have a better understanding of certain topics. It is regrettable that the course goes sometimes too fast (some examples described in detail would be very useful for understanding) and that teachers are not always easy to understand.
by Pavel C•
I'm very happy to accomplish this course! Now I have a much clearer picture of RL methods.
In order to pass this course you'll require a good knowledge of python and some nonzero experience with tensorflow. Some tasks are really hard to pass, once I even had to install environment and run training on my home computer for several hours.
I want to say thanks to course authors and a little suggestion: please add topic about curiosity in RL.
by Roland R•
Topic is very interesting and most of the content is presented in an understandable way. A lot of additional material is presented that helps to deepen your knowledge. The programming assignments also help a lot to grasp what really is going on. Some of the notebooks are a little bit broken (missing RL environments, broken submit scripts, some tasks not clear, ...). But all in all very good.
by Guy K•
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 :)
It was avery good learning experience from some of the birghtest minds in deep reinforcement leaning domain. The examples used were very intuitive and this made the whole learning experience very nice. I wish to learn more in DRL and I'm eagar now to explore more problem statements in this area and work on them. Thanks a lot for everything.
by Jonas B•
Content provides a good - and useful :) - overview of reinforcement learning. The hands-on exercises in the notebooks were the main reason I decided to do the course, and I enjoyed doing them. However, they did contain a lot of errors and broken code,. This would need to be fixed for the course to earn a 5/5.
by Sean H•
Overall very informative and well done course, I highly recommend it. The support in the discussion forums is the main area where it lacks. Sometimes the topics are hard to grasp, so it's really a big help when there is good support in the forums.
by Chun T Y•
Some details are not explained as clear as it can be, maybe there can more reading material to bridge the knowledge gap between course syllabus and intermediate level ML experience. Nice work tho!
by Francesco R•
Wonderful content and super interesting problem assignments, but please fix the bugs in the graders and spend some time to adapt the code to the Coursera platform.
by Roman P•
Four stars only because the notebooks/excercises don't work well. Aside of that, I learned a lot in this class. Thank you!!
this course is helpful to learning Reinforce learning, but with some ambiguous context need a lot time to understand,
by Mark C•
Great instructors and course material, but there are enough bugs in the quizzes and assignments to be annoying.
by Shahram N S•
Overall a good course, But there are many bugs and errors in the programming portion of the course.
by Matthieu G•
+ Great coding assignments : practical and motivating !
- Sometimes the videos could be more clear
by Maxim V•
Awesome course, although some quizzes and programming assignments could be more user friendly.
by Emilio P•
Wonderful course. Just would need a little bit more work on the subtitles.
by Потапчук А А•
It still a beta:(
by Philippe M•
Notebooks are very useful to learn the theory and I certainly learned a lot throughout this course. However, codes in homework often lack of clarity and sometime includes errors and not enough assert checks. The audio is also hard to and understand, the slides are often poor, missing keywords, recaps, organization, sometimes you have 4 dense slides per lessons. Would be great to have better slides as a support of the speech. The questions during videos are really useful. The teaching staff is really reactive and many issues, difficulties have already been solved by the community, which helps a lot for autonomy. So thank you, it was really really useful, but it could have been better and cleaner !
by Kristoffer M•
The teachings are actually quite good, and the problem sets are fine, but there are so many bugs in the submission methods that you spend halv of your coding time trying to debug the submit methods. Frustrating. If they fix the course, this course will be highly recommended.