Bayesian Robust Optimization for Imitation Learning

Abstract

One of the main challenges in imitation learning is determining what action an agent should take when outside the state distribution of the demonstrations. Inverse reinforcement learning (IRL) can enable generalization to new states by learning a parameterized reward function, but these approaches still face uncertainty over the true reward function and corresponding optimal policy. Existing safe imitation learning approaches based on IRL deal with this uncertainty using a maxmin framework that optimizes a policy under the assumption of an adversarial reward function, whereas risk-neutral IRL approaches either optimize a policy for the mean or MAP reward function. While completely ignoring risk can lead to overly aggressive and unsafe policies, optimizing in a fully adversarial sense is also problematic as it can lead to overly conservative policies that perform poorly in practice. To provide a bridge between these two extremes, we propose Bayesian Robust Optimization for Imitation Learning (BROIL). BROIL leverages Bayesian reward function inference and a user specific risk tolerance to efficiently optimize a robust policy that balances expected return and conditional value at risk. Our empirical results show that BROIL provides a natural way to interpolate between return-maximizing and risk-minimizing behaviors and outperforms existing risk-sensitive and risk-neutral inverse reinforcement learning algorithms. Code is available at https://github.com/dsbrown1331/broil.

In proceedings NeurIPS 2020

References in corpus (14)

• A Simple Framework for Contrastive Learning of Visual Representations
• Safe Policy Improvement with Baseline Bootstrapping
• Safe Policy Improvement by Minimizing Robust Baseline Regret
• Decoupling Representation Learning from Reinforcement Learning
• Regret-based Reward Elicitation for Markov Decision Processes
• Safe Imitation Learning via Fast Bayesian Reward Inference from Preferences
• Worst Cases Policy Gradients
• Soft-Robust Actor-Critic Policy-Gradient
• Deep Bayesian Reward Learning from Preferences
• An Approximate Solution Method for Large Risk-Averse Markov Decision Processes
• Risk-Sensitive Generative Adversarial Imitation Learning
• Partial Policy Iteration for L1-Robust Markov Decision Processes
• Learning Safe Policies with Expert Guidance
• Epistemic Risk-Sensitive Reinforcement Learning

Cited by in corpus (2)

• Learning from Ambiguous Demonstrations with Self-Explanation Guided Reinforcement Learning
• Universal Off-Policy Evaluation