Guardado en:
Detalles Bibliográficos
Autores principales: Zhang, Ruiyuan, Chen, Yuyao, Huo, Yuchi, Liu, Jiaxiang, Xi, Dianbing, Liu, Jie, Wu, Chao
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2410.03778
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912060020883456
author Zhang, Ruiyuan
Chen, Yuyao
Huo, Yuchi
Liu, Jiaxiang
Xi, Dianbing
Liu, Jie
Wu, Chao
author_facet Zhang, Ruiyuan
Chen, Yuyao
Huo, Yuchi
Liu, Jiaxiang
Xi, Dianbing
Liu, Jie
Wu, Chao
contents Multi-task-learning(MTL) is a multi-target optimization task. Neural networks try to realize each target using a shared interpretative space within MTL. However, as the scale of datasets expands and the complexity of tasks increases, knowledge sharing becomes increasingly challenging. In this paper, we first re-examine previous cross-attention MTL methods from the perspective of noise. We theoretically analyze this issue and identify it as a flaw in the cross-attention mechanism. To address this issue, we propose an information bottleneck knowledge extraction module (KEM). This module aims to reduce inter-task interference by constraining the flow of information, thereby reducing computational complexity. Furthermore, we have employed neural collapse to stabilize the knowledge-selection process. That is, before input to KEM, we projected the features into ETF space. This mapping makes our method more robust. We implemented and conducted comparative experiments with this method on multiple datasets. The results demonstrate that our approach significantly outperforms existing methods in multi-task learning.
format Preprint
id arxiv_https___arxiv_org_abs_2410_03778
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SGW-based Multi-Task Learning in Vision Tasks
Zhang, Ruiyuan
Chen, Yuyao
Huo, Yuchi
Liu, Jiaxiang
Xi, Dianbing
Liu, Jie
Wu, Chao
Computer Vision and Pattern Recognition
Machine Learning
Multi-task-learning(MTL) is a multi-target optimization task. Neural networks try to realize each target using a shared interpretative space within MTL. However, as the scale of datasets expands and the complexity of tasks increases, knowledge sharing becomes increasingly challenging. In this paper, we first re-examine previous cross-attention MTL methods from the perspective of noise. We theoretically analyze this issue and identify it as a flaw in the cross-attention mechanism. To address this issue, we propose an information bottleneck knowledge extraction module (KEM). This module aims to reduce inter-task interference by constraining the flow of information, thereby reducing computational complexity. Furthermore, we have employed neural collapse to stabilize the knowledge-selection process. That is, before input to KEM, we projected the features into ETF space. This mapping makes our method more robust. We implemented and conducted comparative experiments with this method on multiple datasets. The results demonstrate that our approach significantly outperforms existing methods in multi-task learning.
title SGW-based Multi-Task Learning in Vision Tasks
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2410.03778