Modern Reinforcement Learning: Deep Q Agents (PyTorch & TF2) (Udemy.com)

I do not have a strong mathematical background, so it was hard to follow this part of the course. A little more explanation here would be better. On the other hand, I could follow the coding part, and now I can understand the basic ideas of the course. With the new knowledge, I could achieve results in other environments. Thanks.

I've never had this much fun with a Udemy course. Your voice is sooth and calming, but not to the point that it gets boring. Your ways of explaining things combined with the powerpoints is charming. Also, your ways of teaching feel so inviting and judgement free. It's been my dream to train an AI to play a videogame. Now the dream has become reality. Thank you.

After a while, the complexity increases and the quality of "teaching" decreases... Very disappointing really, the distribution of the course leaves subjects pending so you have to keep watching videos. While you wait for the author to resume the subject, it is already late. What you were waiting for is not explained and you can't refund the course. Example: Section 3... 55 minutes explaining DQL for continuous state spaces and the code doesn't have the implementation of memory, he ends the sections showing bad performance in his code(even he says it). Then in Section 4... He rectifies his own code, saying that the ways implemented before were wrong..... Meanwhile he's coding, he does not explain how this code fit into the theory of the algorithm... Summary: all the explanation is superficial, it's much better if you just check the code of his github and google the subjects or parts of the code, because everything in this course you can get it for free somewhere else. I've seen a lot of videos on youtube with the same or even better quality content. If you are looking for a great explanatory step-by-step code with a nice gradually increase in complexity: go somewhere else, this is not the course. The idea of paying for a course is that the author MUST give you knowledge(quality or insides) that you can't get it for free somewhere else.

Thank you Phil for the great work, really made a complex subject more accessible. I apricated the time and effort that you put to go over the papers and the code in detail. The course delivers on its title and goes over deep Q learning in pytorch with clean oop code. The math part was fast and furious but as Phil mentioned at the end, there is literature out there that goes in-depth if necessary. The only thing I would like to see added to this course is: how to deal with custom environments and how to model/use them and not just the Atari library.

Stumbled upon this course by chance while browsing data science courses. Checked out the preview, and I've immediately recognized the fact that I've stumbled upon gold. Very few, and I repeat, very very few courses out there offer such pro material as this course. No watered down rehearsed silly Jupyter notebooks. No attempt to make it "friendly" and "accessible" to mass audience. This is a course by a pro, for aspiring pros. The length is perfect. The course is practical and right on point. The author doesn't explain the basics, which is perfect. Why should he? This is a course for advanced material. The code is written on Linux (Ubuntu I guess). OOP principles are strictly kept. Honestly, I could go on and on about how much I enjoyed the level of meticulously in this course. Definitely 5 stars. Would give it more if I could. Starting the 2nd course now :)

First: My background as reference Mechanical Engineering Degree I have taken Andrew Ng's and Jeremy Howard's courses before this, and this is my first foray into reinforcement learning. I have been scripting (now getting more into the OOP features of python) in Python and MATLAB for a couple years now with lots of work in neural networks and libraries such as FastAI. However, I am left feeling there are significant gaps between the material covered (lack of explanation sort giving the reasons in human-ese) and the assignments. The prerequisite knowledge Phil covered at the beginning of the course (how well you need to know xyz for this course) is not consistent with the actual course material or assignments. I'm finding myself spending hours pouring over the paper (and others) to get an understanding of WHY we are buffer, memory counter, etc. I'm attempting to use the Feynman method, but at the end of the day I'm having a tough time understanding the fundamentals of the issues behind our Naive-DQN. Correlated sequences of observations being an issue is something I THINK I understand, but not quite the other two. This is just a long-winded way of saying the level Phil is teaching at when explaining these concepts is a bit too high, and the things that appear to be obvious to him are pretty far above the expectations laid out in the beginning of our level. I'm determined to learn this, but it will definitely have to be through re-attacking methods of deconstructing these concepts of the first paper so I can begin to digest them piece-by-piece at a deeper level.

I am not yet familiar enough with python/pytorch to write the code in the exercises. So I am going through the course first and will probably repeat again in the future. The material is perfect for what I want to learn. I think the pseudo code should be put in a pdf resource to make it easier to access, rather than navigate back to find them.

I pushed myself to quickly go through the material, mainly to familiarize myself with the terminology, I have a lot more work to do along with learning Python. I know that it will take me awhile to learn how to go from paper to code but I know that it will be worth while. Thank you again for creating this class. Fred Reiss

The absolutely the deepest deep reinforcement learning course you find anywhere! Phil is a "Deep Learning Machine!" who leads you on an epic journey from the ground up. Phil gets to the point without wasting time, building from a basic blocks of reusable code to q_learning, to Deep_q_learning, to Double_Deep_Q_Learning, to Dueling_Double_Deep_Q_Learning! In addition to that he teaches you how to read scientific papers! The value of this course cannot be overstated. This is a Deep Dive into the Deep Reinforcement Learning!

El desarrollo del código es modular, lo que significa que puedes adaptarlo a tus proyectos solo modificando tu ambiente y tu tipo de red neuronal. Sin embargo, durante el desarrollo del código la explicación no mes muy útil. El curso esta enfocado a estudiantes de maestría y doctorado, con conocimiento previo en programación, redes neuronales y machine learning