NeurIPS 2024poster1 citations
GO4Align: Group Optimization for Multi-Task Alignment
Jiayi Shen, Cheems Wang, Zehao Xiao, Nanne Van Noord, Marcel Worring
Abstract
This paper proposes **GO4Align**, a multi-task optimization approach that tackles task imbalance by explicitly aligning the optimization across tasks. To achieve this, we design an adaptive group risk minimization strategy, comprising two techniques in implementation: (i) dynamical group assignment, which clusters similar tasks based on task interactions; (ii) risk-guided group indicators, which exploit consistent task correlations with risk information from previous iterations. Comprehensive experimental results on diverse benchmarks demonstrate our method's performance superiority with even lower computational costs.
multi-task learningmulti-task optimizationtask groupingdense prediction tasks
BibTeX
@inproceedings{
shen2024goalign,
title={{GO}4Align: Group Optimization for Multi-Task Alignment},
author={Jiayi Shen and Cheems Wang and Zehao Xiao and Nanne Van Noord and Marcel Worring},
booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
year={2024},
url={https://openreview.net/forum?id=8vCs5U9Hbt}
}