Mao Hong

2 accepted papers

2024

A Policy Gradient Method for Confounded POMDPs

ICLR 2024poster

In this paper, we propose a policy gradient method for confounded partially observable Markov decision processes (POMDPs) with continuous state and observation spaces in the offline setting. We first establish a novel identification result to non-parametrically estimate any history-dependent policy…

Cited by 6 Source ↗PDF ↗Scholar ↗

2024

Model-based Reinforcement Learning for Confounded POMDPs

Mao Hong, Zhengling Qi, Yanxun Xu

ICML 2024poster

We propose a model-based offline reinforcement learning (RL) algorithm for confounded partially observable Markov decision processes (POMDPs) under general function approximations and show it is provably efficient under some technical conditions such as the partial coverage imposed on the offline da…

Cited by 1 Source ↗PDF ↗Scholar ↗