This is by far the most comprehensible RL course available online. It does not mean easy but the way instructor take you each concept one at a time makes it easy to grasp the concepts which I think are confusing at times.

This course is one of the best I've learned so far in coursera. The explanations are clear and concise enough. It took a while for me to understand Bellman equation but when I did, it felt amazing!

An excellent introduction to the subject of Reinforcement Learning, accompanied by a very clear text book. The python assignments in Jupyter notebooks are both informative and helpful.

One of the best courses I finished on Coursera, I really like the structure of the course. Textbook is also provided which really helps. Looking forward to next course in the series.

Very good courses. It helps me to understand reinforcement learning a lot.

A great course to start learning Reinforcement Learning!

The course is overall well structured and concise. I am sure that the instructors could do a better job of putting more emphasis on the difficult parts of the course (such as how to actually use the Bellman Equations and how to calculate the State/Action Value functions). More examples of calculation would have made things far easier. All in all, it was a decent introduction to RL and the videos cleared some of the confusion that arised just by reading the RL handbook by Sutton & Barto.

Some of the sections seemed a bit rushed up. While the book provided a good source to clarify, I would prefer a slightly slower pace with emphasis on understanding during the video presentation. However, I have learnt significantly on reinforcement learning during the course. Thanks to the instructors who are highly accomplished, and have taken the time to create this video course.

What ever the content taught was really really good. but still more hands-on algorithms such as Monte Carlo would have been even better. overall worth studying the course

Good overall. Need to work with your assignments and their submission procedure. Lectures should be more interactive than just going through slides.

Not worth it.

If you have solid foundation, this foundation can be better understood by reading textbook for around 2 hours.

If you don't have solid foundation, then you shouldn't touch RL at all.

not enough support to check computer exercices

They put a lot of effort into it the course, however, they choose for some reason not to share the slides with their students. The accompanying book may be the standard, but yet it does not summarize the content as the slides do.

The programming examples are to simple and to few.

A vast amount of the video contains 'what we are going to cover' and 'what we have have'. This would make sense, if there are longer videos, but not if there is just one or two minutes of content.

The course is very dry. It's almost like the instructors are reading the book out loud for you. The videos from other sources they provide seem out of place, like cutting and pasting into the course without proper editing. Furthermore, I don't see the use of seeing instructors' faces during the lecture. It is more distractive than helpful.

the first and only other coursera course I took was mathematics of machine learning from imperial university of london. I found it challenging and educational, with fantastic presentation. it may serve as a good model to improve this course

A Lot of theoretical math and Too few code I recommend to show this complex mathematical equetion in code also

My rating and review is for the entire specialization. Big compliments to the course instructors for developing a truly insightful and challenging course. Out of all the specializations and courses I have done on Coursera, this definitely has been the most challenging and rewarding one. The use of a textbook in addition to the video and programming materials is a very useful approach - to get the most out of this course, I recommend everyone to study all the material in advance of the lectures. This really increases the added value of the course.

As has been said by others, the capstone project (course 4) is the least developed. It is mostly a reiteration of videos from previous courses, and a few programming assignments (numbers 1 and 3 of which are very easy, whereas number 2 is very challenging). A drawback is that as the courses progress, but in particular in course 4, the programming exercises become quite mechanical, and mostly involve following specific instructions with a bit of debugging, without keeping proper sight of the bigger picture. Then again, I've felt this is the case for many courses in ML, and not particular to this course.

At this point, I feel I have a pretty good understanding of the theoretical basis of RL, but I would definitely not be able to implement an RL agent independently. I guess with this basis in hand, it is now time for some more applied (self-)study in this domain. I have definitely become even more eager to do so.

Very well put together course. It does a good job of walking you through concepts in a way that's direct and accessible, while not dumbing things down. I had bought the Intro to RL textbook some months back but ran into problems getting its material to 'stick', but the ideas in chapters 1-4 are much more concrete now.

Assignments were reasonably difficult, but not overwhelmingly so. Homeworks are designed to make sure you understand key concepts moreso than being vigilantly 'rigorous' for their own sake. Emphasis of the class is making sure you understand fundamental concepts moreso than hacking your way into a working prototype of something.

While the class is designed to be easily digested, the material assumes a working knowledge of programming and mathematical formalism, so people without some background knowledge you might struggle to keep pace, even if the material is well designed.

Also, like others have mentioned.......this class follows the book pretty carefully. Don't expect anything to be covered that's not in the RL textbook the course is based off of. By the end of the class the book material will be more vivid and concrete in your mind, but you will not have branched into a direction not covered within it.

Great! My only gripe pertains to week 4's presentation of "dynamic programming". Dynamic programming is a huge beast of a topic on its own, and the book chapter only scratches 0.01 % of the surface. A minor suggestion would be a better overview and example of a typical dynamic programming problem and solution. The course does not prepare you to answer the last discussion question: "what are some applications of dynamic programming". In order to answer this, you will either 1. copying answers off google,.... or 2. Get and learn from a book on dynamic programming. Despite this one minor gripe, all the weekly modules including week 4 does give a lot of valuable information. It's an amazing resource to accompany the RL book by Sutton and Barto, and actually helps solving the problems in the book. The programming question in week 4 is excellent because it they make it pretty easy, where you fill in code for a few functions. Most of the code is already coded by the course experts and all the files can be downloaded and you can analyze and learn the details of implementation on your own time. Thank you for the great resource!

The course is very good, I already had some experience with Markov chains. I found the hardest part to understand was week 3 with the Bellman equations. The course should be reinforced in this part so that everyone can understand, and it would even be to pass some of the material in this section to the second week (since it is a lot of topic).

I had to do a lot of research with other sources to be able to understand the content of week 3, since the book is not very clear material, especially on this topic. It would be very interesting if you could explain us better how to create these environments that are mentioned (eg GridWorld, Pole Car, etc) in Python in order to improve the form and intuition in the application of these concepts, this as off-topic material.

The guest talks were very beneficial to me, they give some very interesting historical and practical perspectives on the application of the concepts. Professor Warren Powell's talk was the most interesting in my opinion and he left me wondering how we could apply these concepts to real life problems.

Thanks to the team for this course.

The same review for all 4 courses: This is the first time I have done a Coursera module building courses up rather than just individual courses. You really feel you have achieved something out of it. Some people have commented that it is just presenting material from the Sutton and Barto book. But that book is *the* text book in the subject. The course selected particular chapters from the book. I wouldn't have got as much just from trying to read the book on its own (I probably wouldn't have read as much as I did). It was good to have the supplementary videos with other experts - and great to watch Sutton and Barto just sat down being recorded having a retrospective conversation. The programming exercises would sometimes feel they weren't testing much (in fact, the challenges were largely due to my lack of skill in Python - my Python abilities have improved which is a side benefit) but they would actually get you into the weeds as it were. All in all, the best courses I've done. Great job Martha and Adam!

After being laid off from my big-tech job I decided to use that opportunity to learn about new areas in machine learning. I had long wanted to explore the field of reinforcement learning. Reinforcement Learning is one of the most beautiful discoveries in Machine Learning. It's fascinating to see how a simple idea such as the reward hypothesis is so powerful in practice. Hearing the dialogue between Rich Sutton and Andrew Barto was very inspiring. I look forward to taking more courses in this specialization and learning more about RL. I would also like to request the instructors to create a course on the more recent trends in RL such as Deep Reinforcement Learning. There isn't a very good resource online to learn about Deep RL methods and reproducing them reliably.

I think it's really good course, probably the best that I have done on Coursera (and I have done a few). I like the approach of combining independent reading with high-quality video content. They do a great job of breaking down complicated topics in reinforcement learning, and the programming problems help gain a deeper and intuitive understanding of the underlying maths. I think some of the proofs in the book could be more detailed. For one or two I went on line to find more step-by-step proofs. This is in principle more an issue with the book and less with the course. But I think it could be helpful if the course would provide a reference to some more in-detail proofs. That's it though, I am really enjoying this specialization :)

I recently completed the "Fundamentals of Reinforcement Learning" course on Coursera, and I couldn't be happier with the experience. This course exceeded my expectations in every way, offering a comprehensive exploration of reinforcement learning. The content was well-structured. The lecturers were knowledgeable and passionate, delivering engaging lectures with easy to understand examples. The teaching methods, including interactive quizzes and assignments, were effective in reinforcing the concepts. While the programming assignments presented a challenge, they provided valuable hands-on experience, even if they involved completing code blanks. Overall, I highly recommend this course to anyone interested in reinforcement learning.

This course teaches you by having you read the textbook chapters (free pdf) followed by complementary videos that help you gain intuition. The quizzes focus on you truly understanding the material and are not easy. Quizzes plus two programming assignments help you actively learn rather than just passively reading or watching videos. I do wish the Discussion Forums kept a longer history of Q&A and had more responses from the instructors and mentors. For example, I had a question that someone else had already asked (so it wasn't an unusual question), but there was no response, and enough time had passed where the post became locked from any responses. So, I would really rate this course as 4.5 stars.