Oct 23, 2023·edited Feb 8Liked by Cameron R. Wolfe, Ph.D.
Cameron, could you recommend a SOTA textbook aimed at a graduate level audience on this topic? I know it is a dynamic area and a moving target, but a foundational text would be great.
This page is a sort of text book for deep RL and (by far) the most comprehensive resource in terms of aggregating, explaining, and applying modern RL algorithms. They have links to several useful sources, including papers for each notable algorithm and a variety of textbooks. Two texts that are referenced frequently are:
- Sutton, Richard S., et al. "Policy gradient methods for reinforcement learning with function approximation." Advances in neural information processing systems 12 (1999).
However, these obviously don't include the most recent algorithms due to being written in the 90s.
Cameron, could you recommend a SOTA textbook aimed at a graduate level audience on this topic? I know it is a dynamic area and a moving target, but a foundational text would be great.
Great question! My recommendation would be to check out this super useful webpage from OpenAI: https://spinningup.openai.com/en/latest/
This page is a sort of text book for deep RL and (by far) the most comprehensive resource in terms of aggregating, explaining, and applying modern RL algorithms. They have links to several useful sources, including papers for each notable algorithm and a variety of textbooks. Two texts that are referenced frequently are:
- Simple statistical gradient-following algorithms for connectionist reinforcement learning, Williams, Machine learning 1992
- Sutton, Richard S., et al. "Policy gradient methods for reinforcement learning with function approximation." Advances in neural information processing systems 12 (1999).
However, these obviously don't include the most recent algorithms due to being written in the 90s.