Confident Off-policy Evaluation and Selection through Self-Normalized Importance Weighting

Channel:
United States Simons Institute for the Theory of Computing
Subscribers:
68,700
Published on ● Video Link: https://www.youtube.com/watch?v=0MYRwW6BdvU


Duration: 41:01
787 views
10

Ilja Kuzborskij (DeepMind)
https://simons.berkeley.edu/talks/tbd-238
Reinforcement Learning from Batch Data and Simulation



Other Videos By Simons Institute for the Theory of Computing


2020-12-03Panel Discussion
2020-12-02The Mean-Squared Error of Double Q-Learning
2020-12-02Q-learning with Uniformly Bounded Variance
2020-12-02Zap Stochastic Approximation and Implications to Q-Learning
2020-12-02Computational/Statistical Gaps for Learning Neural Networks
2020-12-02Uniform Offline Policy Evaluation (OPE) and Offline Learning in Tabular RL
2020-12-02Batch Value-function Approximation with Only Realizability
2020-12-01Reinforcement Learning using Generative Models for Continuous State and Action Space Systems
2020-12-01Monte Carlo Sampling Approach to Solving Stochastic Multistage Programs
2020-12-01Robust Learning of Stochastic Dynamical Systems
2020-12-01Confident Off-policy Evaluation and Selection through Self-Normalized Importance Weighting
2020-12-01An Equivalence between Loss Functions and Non-Uniform Sampling in Experience Replay
2020-11-30Beyond Worst-Case: Instance-Dependent Optimality in Reinforcement Learning
2020-11-30Learning Multi-Agent Collaborations With Decomposition
2020-11-30Online Learning with A Lot of Batch Data
2020-11-24Ahmed El Alaoui | Fellows Lightning Talks | 6th Annual Industry Day
2020-11-24Computational Complexity of Statistical Inference | Program Presentations | 6th Annual Industry Day
2020-11-24Computational Innovation and Data-Driven Biology | Program Presentations | 6th Annual Industry Day
2020-11-24JP Morgan | Industry Partner Lightning Talks | 6th Annual Industry Day
2020-11-24Microsoft Research | Industry Partner Lightning Talks | 6th Annual Industry Day
2020-11-24Vidya Muthukumar | Fellows Lightning Talks | 6th Annual Industry Day


Tags:
Simons Institute
theoretical computer science
UC Berkeley
Computer Science
Theory of Computation
Theory of Computing
Reinforcement Learning from Batch Data and Simulation
Ilja Kuzborskij