Deep PQR: Solving Inverse Reinforcement Learning using Anchor Actions

Abstract

We propose a reward function estimation framework for inverse reinforcement learning with deep energy-based policies. We name our method PQR, as it sequentially estimates the Policy, the $Q$-function, and the Reward function by deep learning. PQR does not assume that the reward solely depends on the state, instead it allows for a dependency on the choice of action. Moreover, PQR allows for stochastic state transitions. To accomplish this, we assume the existence of one anchor action whose reward is known, typically the action of doing nothing, yielding no reward. We present both estimators and algorithms for the PQR method. When the environment transition is known, we prove that the PQR reward estimator uniquely recovers the true reward. With unknown transitions, we bound the estimation error of PQR. Finally, the performance of PQR is demonstrated by synthetic and real-world datasets.

References in corpus (8)

• Guided Cost Learning: Deep Inverse Optimal Control via Policy Optimization
• Reinforcement Learning with Deep Energy-Based Policies
• Path integrals and symmetry breaking for optimal control theory
• Maximum Entropy Deep Inverse Reinforcement Learning
• Causal Confusion in Imitation Learning
• Scalable agent alignment via reward modeling: a research direction
• Ivy: Instrumental Variable Synthesis for Causal Inference
• An Efficient Pseudo-likelihood Method for Sparse Binary Pairwise Markov Network Estimation