Aligned Multi Objective Optimization

Yonathan Efroni, Ben Kretzu, Daniel R. Jiang, Jalaj Bhandari, Zheqing Zhu, Karen Ullrich
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:14989-15017, 2025.

Abstract

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 its superior performance compared to naive approaches.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-efroni25a, title = {Aligned Multi Objective Optimization}, author = {Efroni, Yonathan and Kretzu, Ben and Jiang, Daniel R. and Bhandari, Jalaj and Zhu, Zheqing and Ullrich, Karen}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {14989--15017}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/efroni25a/efroni25a.pdf}, url = {https://proceedings.mlr.press/v267/efroni25a.html}, abstract = {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 its superior performance compared to naive approaches.} }
Endnote
%0 Conference Paper %T Aligned Multi Objective Optimization %A Yonathan Efroni %A Ben Kretzu %A Daniel R. Jiang %A Jalaj Bhandari %A Zheqing Zhu %A Karen Ullrich %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-efroni25a %I PMLR %P 14989--15017 %U https://proceedings.mlr.press/v267/efroni25a.html %V 267 %X 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 its superior performance compared to naive approaches.
APA
Efroni, Y., Kretzu, B., Jiang, D.R., Bhandari, J., Zhu, Z. & Ullrich, K.. (2025). Aligned Multi Objective Optimization. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:14989-15017 Available from https://proceedings.mlr.press/v267/efroni25a.html.

Related Material