Pareto Set Learning for Multi-Objective Reinforcement Learning

TL;DR

This paper introduces PSL-MORL, a novel framework that uses hypernetworks to generate diverse policies for multi-objective reinforcement learning, effectively covering the Pareto front and outperforming existing methods.

Contribution

The paper presents a decomposition-based hypernetwork framework for MORL that produces personalized policies for different preferences, enhancing Pareto front coverage and efficiency.

Findings

Achieves dense Pareto front coverage in experiments.

Outperforms state-of-the-art MORL methods in hypervolume.

Demonstrates theoretical guarantees of model capacity and policy optimality.

Abstract

Multi-objective decision-making problems have emerged in numerous real-world scenarios, such as video games, navigation and robotics. Considering the clear advantages of Reinforcement Learning (RL) in optimizing decision-making processes, researchers have delved into the development of Multi-Objective RL (MORL) methods for solving multi-objective decision problems. However, previous methods either cannot obtain the entire Pareto front, or employ only a single policy network for all the preferences over multiple objectives, which may not produce personalized solutions for each preference. To address these limitations, we propose a novel decomposition-based framework for MORL, Pareto Set Learning for MORL (PSL-MORL), that harnesses the generation capability of hypernetwork to produce the parameters of the policy network for each decomposition weight, generating relatively distinct policies for various scalarized subproblems with high efficiency. PSL-MORL is a general framework, which is compatible for any RL algorithm. The theoretical result guarantees the superiority of the model capacity of PSL-MORL and the optimality of the obtained policy network. Through extensive experiments on diverse benchmarks, we demonstrate the effectiveness of PSL-MORL in achieving dense coverage of the Pareto front, significantly outperforming state-of-the-art MORL methods in the hypervolume and sparsity indicators.