2025 ICML ICML 2025

Efficient First-Order Optimization on the Pareto Set for Multi-Objective Learning under Preference Guidance