提出面向相关目标的优化框架，实现多任务性能协同提升。
Aligned Multi Objective Optimization
- 基于目标对齐设计新型梯度优化方法
- 在多任务学习中实现各目标同时增益
- 适合需协同优化多个相关任务的场景
当前多目标优化研究多聚焦于冲突目标，关注帕累托前沿或用户权衡。然而，在机器学习实践中，存在大量非冲突场景：多任务学习、强化学习与大模型训练表明，相关任务可同时提升各项性能。尽管已有证据支持这一现象，但缺乏从优化角度的系统研究。这导致缺少可扩展至大量相关目标的通用梯度方法。为此，本文提出对齐多目标优化框架，设计新算法，并提供理论证明其优于朴素方法。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。