Learning to Be Cautious

TL;DR

This paper introduces a reinforcement learning algorithm that enables agents to learn cautious behaviors in novel situations by modeling reward uncertainty with neural network ensembles, eliminating the need for task-specific safety tuning.

Contribution

It presents a novel approach that learns cautious policies through reward function uncertainty without relying on explicit safety information or task-specific tuning.

Findings

Agents learn cautious behaviors in complex tasks.

The approach outperforms safety-tuned baselines.

Caution emerges without explicit safety constraints.

Abstract

A key challenge in the field of reinforcement learning is to develop agents that behave cautiously in novel situations. It is generally impossible to anticipate all situations that an autonomous system may face or what behavior would best avoid bad outcomes. An agent that can learn to be cautious would overcome this challenge by discovering for itself when and how to behave cautiously. In contrast, current approaches typically embed task-specific safety information or explicit cautious behaviors into the system, which is error-prone and imposes extra burdens on practitioners. In this paper, we present both a sequence of tasks where cautious behavior becomes increasingly non-obvious, as well as an algorithm to demonstrate that it is possible for a system to learn to be cautious. The essential features of our algorithm are that it characterizes reward function uncertainty without task-specific safety information and uses this uncertainty to construct a robust policy. Specifically, we construct robust policies with a k-of-N counterfactual regret minimization (CFR) subroutine given learned reward function uncertainty represented by a neural network ensemble. These policies exhibit caution in each of our tasks without any task-specific safety tuning. Our code is available at https://github.com/montaserFath/Learning-to-be-Cautious