---
title: "Injecting Imbalance Sensitivity for Multi-Task Learning"
authors:
  - "Zhipeng Zhou"
  - "Liu Liu"
  - "Peilin Zhao"
  - "Wei Gong"
doi: "10.24963/ijcai.2025/805"
url: "https://ijcai.org/proceedings/2025/805"
pdf: "https://ijcai.org/proceedings/2025/0805.pdf"
venue: "Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence"
year: 2025
pages: "7236-7244"
track: "Main Track"
keywords:
  - "Machine Learning: ML: Multi-task and transfer learning"
  - "Machine Learning: ML: Meta-learning"
---

# Injecting Imbalance Sensitivity for Multi-Task Learning

Zhipeng Zhou, Liu Liu, Peilin Zhao, Wei Gong

*Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence*, pages 7236-7244. DOI: [10.24963/ijcai.2025/805](https://doi.org/10.24963/ijcai.2025/805)

## Abstract

Multi-task learning (MTL) has emerged as a promising approach for deploying deep learning models in real-life applications. Recent studies have proposed optimization-based learning paradigms to establish task-shared representations in MTL. However, our paper empirically argues that these studies, specifically gradient-based ones, primarily emphasize the conflict issue while neglecting the potentially more significant impact of imbalance/dominance in MTL. In line with this perspective, we enhance the existing baseline method by injecting imbalance-sensitivity through the imposition of constraints on the projected norms. To demonstrate the effectiveness of our proposed IMbalance-sensitive Gradient (IMGrad) descent method, we evaluate it on multiple mainstream MTL benchmarks, encompassing supervised learning tasks as well as reinforcement learning. The experimental results consistently demonstrate competitive performance.