This repository contains code for ''A Model-agnostic Approach to Mitigate Gradient Interference for Multi-task Learning".
The CIFAR-100-MTL can be found here.
The CelebA can be found here.
The MultiMNIST can be found here.
The CityScapes can be found here.
The NYUv2 can be found here.
The Taskonomy can be found here.
If you find our approach useful in your research, please consider citing:
@article{chai2022model,
title={A model-agnostic approach to mitigate gradient interference for multi-task learning},
author={Chai, Heyan and Yin, Zhe and Ding, Ye and Liu, Li and Fang, Binxing and Liao, Qing},
journal={IEEE Transactions on Cybernetics},
year={2022},
publisher={IEEE}
}