David Silver's lecture series on reinforcement learning, recorded at UCL and tied to Sutton and Barto's textbook, covering MDPs, dynamic programming, Monte-Carlo and temporal-difference methods, and function approximation. The lectures are freely available on YouTube.

Commenters repeatedly call it a very good, excellent introduction to RL that pairs well with Sutton and Barto, and suggest it as the starting point for beginners.

On Reddit (17)

1. Strongly recommendsAacronr/MachineLearning2020in “[D] Kaggle features RL competition with Google Football and Manchester City F.C.”▲ 1

   David Silvers class is excellent and covers the material in the book, they can be done sequentially or concurrently. I did Emma Brunskill's course which I can recommend equally. Once you're through those there's another book I sadly can't remember the name of (I haven't read it yet) that does a…

2. Strongly recommendsalexhuhcyar/MachineLearning2019in “[D] What are your favorite videos / lectures on advanced topics in machine learning?”▲ 3

   David Silver's RL Course is really good: https://www.youtube.com/watch?v=2pWv7GOvuf0

3. Strongly recommendsNoumenon1849r/MachineLearning2017in “[D] How was the deep RL boot camp from deepbootcamp.io”

   I had some prior experience with RL and I found it a little difficult to follow through some of the presentations. I recommend starting with David Silver's course if you're totally new. https://www.youtube.com/watch?v=2pWv7GOvuf0&list=PLzuuYNsE1EZAXYR4FJ75jcJseBmo4KQ9- As a minimum, you should…

4. Strongly recommendstod315r/MachineLearning2015in “[Help] What are the prerequisites for Reinforcement Learning and what are some good resources to get started?”

   I would also highly suggest David Silver's lectures at UCL https://www.youtube.com/watch?v=2pWv7GOvuf0

5. Recommendsfkdosilovicr/MachineLearning2020in “[D] Reinforcement learning online courses”▲ 2

   To be honest, even though David Silver's course is really good, my approach was to read a chapter from Sutton and Barto's book first and then watch the corresponding lecture. After a few chapters, I realized that lectures did not bring anything new, so I stopped watching them. But that was only…

6. Recommends140doritosr/learnmachinelearning2020in “Sutton book + Silver lectures - How to combine them?”▲ 1

   I'm following the Reinforcement Learning Specialization from University of Alberta on Coursera. It is a great course and it basically follows the book and in each chapter it specifies the page numbers that covered in that chapter. David Silver's lectures are very similiar to lectures on this…

7. RecommendsTaxtro1r/learnmachinelearning2020in “How to obtain multiple solutions using RL or NNs”▲ 1

   A) Sounds like you want to know the utility of each state. If you are interested in Reinforcement Learning, I recommend David Silver's lecture on youtube. B) From what I understand you want the network to map one number to a pattern of numbers. That should fundamentally be possible. Perhaps try a…

8. Recommendsmetacurser/MachineLearning2019in “[D] Siraj Raval - Potentially exploiting students, banning students asking for refund. Thoughts?”▲ 14

   I don't want to sound cocky, if you were more experienced, you would perhaps see how "flashy" it is. The way he lays out stuff seems to be more about letting others know he knows, rather than explaining things in detail. I would recommend Hinton's or Silver's courses for you to get a deeper and…

9. Recommendsakabertor/MachineLearning2019in “[D] MIT Deep Learning Basics: Introduction and Overview”▲ 5

   Lol. I think this is a problem. There are tons of other folks who contribute to this course (not to mention the actual algorithm). Somehow, plastering his face all over the course somehow seems very disingenuous. Andrew Ng doesn't plaster his face on the Coursera one and that guy is a true pioneer…

10. RecommendsBigMakondor/artificial2017in “OpenAI have developed a bot capable of defeating pro Dota 2 players.”

    You have the gym framework from OpenAI and the newly released [SC2 API from DeepMind.]( https://deepmind.com/blog/deepmind-and-blizzard-open-starcraft-ii-ai-research-environment/) For a more formal education on reinforcement learning you have the David Silver course on video and slides. And also…

11. RecommendsHiestaar/artificial2017in “Enabling Continual Learning in Neural Networks | DeepMind”

    Did you ever think that deep mind objectives are not to reproduce a artificial human brain nor to achieve GAI / singularity? They are specialized in reinforcement learning, a branch of AI focusing on learning how to solve a problem by trial and error. They know a big deal of what they are taking…

12. Recommendsinternet_hamr/MachineLearning2016in “Robotics courses for a Deep Learning guy?”

    In a sense they are inherently highly linked, given that Deep Learning <> AI <> Robotics Even fields like NLP are highly relevant to Human-Robot Interaction. Basically robotics and deep learning are wide branching fields that intersect an awful lot. Because of this I wouldn't really say there are…

13. Recommendsrcsordasr/MachineLearning2016in “Sutton's Reinforcement Learning”▲ 5

    These excellent lectures from David Silver (DeepMind) are also using it as reference: https://www.youtube.com/playlist?list=PL5X3mDkKaJrL42i_jhE4N-p6E2Ol62Ofa

14. Recommendsjlxur/MachineLearning2016in “How to learn about AlphaGo”▲ 2

    Definitely agree with what's been said: - You'll want good background knowledge on RL (David Silver has very nice lectures, and the Sutton & Barto textbook is the foundational text to read). - You'll want good background knowledge on deep learning (Hinton's Coursera course is very good) - I'd also…

15. Recommendsford_beeblebroxr/MachineLearning2015in “"Simple Questions Thread" - 20150708”▲ 2

    Deep Reinforcement for Robotics shows huge potential. --- Deepmind Atari replication is only a few clicks away... Soumith's CVPR 2015 workshop has an Amazon EC2 machine image linked (use a GPU instance) It has torch + itorch + Atari + notebooks (the AMI ID is: ami-b36981d8) all ready to use and a…

16. Recommendschangingourworldr/MachineLearning2015in “Feeding a game map into a neural network”

    Definitely a task for Reinforcement Learning. Take a look at Berkeley AI course, start at Markov Decision Processes also available on edx with more MOOC like feel. David Silver, who works at DeepMind, has an excellent series of lectures. You might also find DeepMind's latest paper about playing…

17. MentionsDevelopigr/learnmachinelearning2020in “What would be the next step after a few introduction classes to RL &amp; AI?”▲ 1

    My best advice would be to spend some time thinking of projects that you think would be really interesting to solve using ML & do them - when you come across something you don’t know how to solve, watch a few YouTube videos on the subject and figure out a way to do it. Maybe the project involved…

Related courses

• Machine Learning (Andrew Ng)Coursera · 240 recommenders
• Neural Networks for Machine Learning (Geoffrey Hinton)Coursera · 25 recommenders
• Reinforcement Learning SpecializationCoursera · 4 recommenders
• Deep Reinforcement Learning (CS 285)UC Berkeley · 2 recommenders
• CS188 Introduction to Artificial IntelligenceUC Berkeley · 2 recommenders
• Practical Deep Learning for Codersfast.ai · 112 recommenders
• CS231n: Convolutional Neural Networks for Visual RecognitionStanford University · 39 recommenders
• Deep Learning Specialization (Andrew Ng)Coursera · 31 recommenders