Opening Challenges and Open Problems in Deep Reinforcement Learning (CS 285, Lecture 23, Part 1)
Challenges and Open Problems in Deep Reinforcement Learning (CS 285, Lecture 23, Part 1)
- Length
- 28 min
- Level
- Undergraduate
- Recorded
- UC Berkeley2023
Chapter 1 of 5: Before the lecture
Examines open problems in deep reinforcement learning, analyzing algorithmic stability and efficiency.
UC Berkeley’s undergraduate CS 285 reviews why deep value-based methods typically lack convergence guarantees. Levine contrasts high variance in policy gradient estimates with the risk that policies exploit errors in learned models.
UC Berkeley’s undergraduate CS 285 reviews why deep value-based methods typically lack convergence guarantees. Levine contrasts high variance in policy gradient estimates with the risk that policies exploit errors in learned models.
DRQ
Evolution Strategies as a Scalable Alternative to Reinforcement Learning
A3C paper

This lecture
Lecture 70 of 73 of Deep Reinforcement Learning.
Choose a node to see how it connects to this lecture. What comes before is on the leftat the top. What comes after is on the rightat the bottom. The ring is what it teaches. People and places are above. Other lectures and books are below.
Press a chosen node again to open it.
Your comment is kept while you sign up.Sign in
TRPO+GAE
Gu et al. '16
Deep Reinforcement Learning in a Handful of Trials
Trust Region Policy Optimization
Human-level control through deep reinforcement learning
Learning to Select and Generalize Striking Movements in Robot Table Tennis
Learning with latent language
Deep reinforcement learning from human preferences