MOPO: Model-Based Offline Policy Optimization

Channel:
United States Simons Institute for the Theory of Computing
Subscribers:
68,700
Published on ● Video Link: https://www.youtube.com/watch?v=hTwsxSd0AxU


Duration: 37:44
2,360 views
39

Tengyu Ma (Stanford
https://simons.berkeley.edu/talks/tbd-206
Deep Reinforcement Learning



Other Videos By Simons Institute for the Theory of Computing


2020-09-30On Distance Approximation for Graph Properties
2020-09-29Discussion: Exploration
2020-09-29PC-PG: Policy Cover Directed Exploration for Provable Policy Gradient Learning
2020-09-29Temporally-Extended ε-Greedy Exploration
2020-09-29Learning Exploration Strategies with Meta-Reinforcement Learning
2020-09-29Behaviour Suite for Reinforcement Learning
2020-09-28Discussion: Offline Reinforcement Learning
2020-09-28Attacking the Off-Policy Problem With Duality
2020-09-28Offline Deep Reinforcement Learning Algorithms
2020-09-28Learning from the Past Without Great Exploration
2020-09-28MOPO: Model-Based Offline Policy Optimization
2020-09-25Pedagogical Talk: Frameworks for Information-Computation Tradeoffs
2020-09-25Reducibility and Statistical-Computational Gaps from Secret Leakage
2020-09-25Berkeley in the 80s, Episode 4: Andrew Yao
2020-09-25Free Energy Wells and Overlap Gap Property in Sparse PCA
2020-09-25The Overlap Gap Property in Principal Submatrix Recovery
2020-09-24Phase Retrieval in High Dimensions: Statistical and Computational Phase Transitions
2020-09-24Pedagogical Talk: Failure of Low-Degree Polynomials
2020-09-24The Rank of Sparse Random Matrices
2020-09-24Sampling From the Random Cluster Model at All Temperatures on Random Graphs
2020-09-24Correlation Decay for Hard Spheres via Markov Chains


Tags:
Tengyu Ma
Simons Institute
Theory of Computing
Theory of Computation
Theoretical Computer Science
Computer Science
UC Berkeley
Deep Reinforcement Learning