Timezone: »

Covariate Shift of Latent Confounders in Imitation and Reinforcement Learning
Guy Tennenholtz · Assaf Hallak · Gal Dalal · Shie Mannor · Gal Chechik · Uri Shalit
Event URL: https://openreview.net/forum?id=phvLhXbBfMM »

We consider the problem of using expert data with unobserved confounders for imitation and reinforcement learning. We begin by defining the problem of learning from confounded expert data in a contextual MDP setup. We analyze the limitations of learning from such data with and without external reward and propose an adjustment of standard imitation learning algorithms to fit this setup. In addition, we discuss the problem of distribution shift between the expert data and the online environment when partial observability is present in the data. We prove possibility and impossibility results for imitation learning under arbitrary distribution shift of the missing covariates. When additional external reward is provided, we propose a sampling procedure that addresses the unknown shift and prove convergence to an optimal solution. Finally, we validate our claims empirically on challenging assistive healthcare and recommender system simulation tasks.

Author Information

Guy Tennenholtz (Technion, Technion)
Assaf Hallak (The Technion)
Gal Dalal (NVIDIA)
Shie Mannor (Technion)
Gal Chechik (NVIDIA, BIU)
Uri Shalit (Technion)

More from the Same Authors