Aligned Multi Objective Optimization
This addresses a gap in optimization for machine learning practitioners dealing with many related objectives, though it appears incremental as it builds on known phenomena from multi-task learning.
The paper tackles the lack of gradient-based methods for multi-objective optimization in scenarios where objectives are aligned rather than conflicting, introducing a new framework with algorithms and theoretical guarantees for superior performance.
To date, the multi-objective optimization literature has mainly focused on conflicting objectives, studying the Pareto front, or requiring users to balance tradeoffs. Yet, in machine learning practice, there are many scenarios where such conflict does not take place. Recent findings from multi-task learning, reinforcement learning, and LLMs training show that diverse related tasks can enhance performance across objectives simultaneously. Despite this evidence, such phenomenon has not been examined from an optimization perspective. This leads to a lack of generic gradient-based methods that can scale to scenarios with a large number of related objectives. To address this gap, we introduce the Aligned Multi-Objective Optimization framework, propose new algorithms for this setting, and provide theoretical guarantees of their superior performance compared to naive approaches.