A Generalized Algorithm for Multi-Objective Reinforcement Learning and Policy Adaptation

TL;DR

This paper presents a generalized algorithm for multi-objective reinforcement learning that enables quick adaptation to new preferences by learning a single policy representation, demonstrated across multiple domains.

Contribution

The authors introduce a generalized Bellman equation for MORL, allowing a single model to adapt efficiently to various preferences and infer preferences with minimal samples.

Findings

Effective adaptation to new preferences with few samples

Successful application across four different domains

Single parametric policy representation for all preferences

Abstract

We introduce a new algorithm for multi-objective reinforcement learning (MORL) with linear preferences, with the goal of enabling few-shot adaptation to new tasks. In MORL, the aim is to learn policies over multiple competing objectives whose relative importance (preferences) is unknown to the agent. While this alleviates dependence on scalar reward design, the expected return of a policy can change significantly with varying preferences, making it challenging to learn a single model to produce optimal policies under different preference conditions. We propose a generalized version of the Bellman equation to learn a single parametric representation for optimal policies over the space of all possible preferences. After an initial learning phase, our agent can execute the optimal policy under any given preference, or automatically infer an underlying preference with very few samples. Experiments across four different domains demonstrate the effectiveness of our approach.