Transforming Vision Transformer: Towards Efficient Multi-Task Asynchronous Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhong, Hanwen, Chen, Jiaxin, Zhang, Yutong, Huang, Di, Wang, Yunhong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910781888528384
author Zhong, Hanwen
Chen, Jiaxin
Zhang, Yutong
Huang, Di
Wang, Yunhong
author_facet Zhong, Hanwen
Chen, Jiaxin
Zhang, Yutong
Huang, Di
Wang, Yunhong
contents Multi-Task Learning (MTL) for Vision Transformer aims at enhancing the model capability by tackling multiple tasks simultaneously. Most recent works have predominantly focused on designing Mixture-of-Experts (MoE) structures and in tegrating Low-Rank Adaptation (LoRA) to efficiently perform multi-task learning. However, their rigid combination hampers both the optimization of MoE and the ef fectiveness of reparameterization of LoRA, leading to sub-optimal performance and low inference speed. In this work, we propose a novel approach dubbed Efficient Multi-Task Learning (EMTAL) by transforming a pre-trained Vision Transformer into an efficient multi-task learner during training, and reparameterizing the learned structure for efficient inference. Specifically, we firstly develop the MoEfied LoRA structure, which decomposes the pre-trained Transformer into a low-rank MoE structure and employ LoRA to fine-tune the parameters. Subsequently, we take into account the intrinsic asynchronous nature of multi-task learning and devise a learning Quality Retaining (QR) optimization mechanism, by leveraging the historical high-quality class logits to prevent a well-trained task from performance degradation. Finally, we design a router fading strategy to integrate the learned parameters into the original Transformer, archiving efficient inference. Extensive experiments on public benchmarks demonstrate the superiority of our method, compared to the state-of-the-art multi-task learning approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2501_06884
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Transforming Vision Transformer: Towards Efficient Multi-Task Asynchronous Learning
Zhong, Hanwen
Chen, Jiaxin
Zhang, Yutong
Huang, Di
Wang, Yunhong
Computer Vision and Pattern Recognition
Multi-Task Learning (MTL) for Vision Transformer aims at enhancing the model capability by tackling multiple tasks simultaneously. Most recent works have predominantly focused on designing Mixture-of-Experts (MoE) structures and in tegrating Low-Rank Adaptation (LoRA) to efficiently perform multi-task learning. However, their rigid combination hampers both the optimization of MoE and the ef fectiveness of reparameterization of LoRA, leading to sub-optimal performance and low inference speed. In this work, we propose a novel approach dubbed Efficient Multi-Task Learning (EMTAL) by transforming a pre-trained Vision Transformer into an efficient multi-task learner during training, and reparameterizing the learned structure for efficient inference. Specifically, we firstly develop the MoEfied LoRA structure, which decomposes the pre-trained Transformer into a low-rank MoE structure and employ LoRA to fine-tune the parameters. Subsequently, we take into account the intrinsic asynchronous nature of multi-task learning and devise a learning Quality Retaining (QR) optimization mechanism, by leveraging the historical high-quality class logits to prevent a well-trained task from performance degradation. Finally, we design a router fading strategy to integrate the learned parameters into the original Transformer, archiving efficient inference. Extensive experiments on public benchmarks demonstrate the superiority of our method, compared to the state-of-the-art multi-task learning approaches.
title Transforming Vision Transformer: Towards Efficient Multi-Task Asynchronous Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2501.06884