Pareto Set Learning for Multi-Objective Reinforcement Learning
This addresses the need for efficient and personalized solutions in multi-objective decision-making for applications like video games, navigation, and robotics, representing a novel method for a known bottleneck.
The paper tackles the problem of generating personalized policies for multi-objective reinforcement learning by proposing PSL-MORL, a decomposition-based framework that uses a hypernetwork to produce policy parameters for each preference, achieving dense coverage of the Pareto front and outperforming state-of-the-art methods in hypervolume and sparsity indicators.
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.