Future-Dependent Value-Based Off-Policy Evaluation in POMDPs.
Masatoshi UeharaHaruka KiyoharaAndrew BennettVictor ChernozhukovNan JiangNathan KallusChengchun ShiWen SunPublished in: NeurIPS (2023)
Keyphrases
- policy evaluation
- reinforcement learning
- partially observable markov decision processes
- markov decision processes
- least squares
- temporal difference
- model free
- policy iteration
- policy gradient
- function approximation
- dynamic programming
- learning algorithm
- belief state
- finite state
- variance reduction
- decision problems
- monte carlo
- markov decision problems
- multi agent
- bayesian networks