Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shi, Jiang-Xin, Wei, Tong, Zhou, Zhi, Shao, Jie-Jing, Han, Xin-Yan, Li, Yu-Feng
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911897654132736
author Shi, Jiang-Xin
Wei, Tong
Zhou, Zhi
Shao, Jie-Jing
Han, Xin-Yan
Li, Yu-Feng
author_facet Shi, Jiang-Xin
Wei, Tong
Zhou, Zhi
Shao, Jie-Jing
Han, Xin-Yan
Li, Yu-Feng
contents The fine-tuning paradigm in addressing long-tail learning tasks has sparked significant interest since the emergence of foundation models. Nonetheless, how fine-tuning impacts performance in long-tail learning was not explicitly quantified. In this paper, we disclose that heavy fine-tuning may even lead to non-negligible performance deterioration on tail classes, and lightweight fine-tuning is more effective. The reason is attributed to inconsistent class conditions caused by heavy fine-tuning. With the observation above, we develop a low-complexity and accurate long-tail learning algorithms LIFT with the goal of facilitating fast prediction and compact models by adaptive lightweight fine-tuning. Experiments clearly verify that both the training time and the learned parameters are significantly reduced with more accurate predictive performance compared with state-of-the-art approaches. The implementation code is available at https://github.com/shijxcs/LIFT.
format Preprint
id arxiv_https___arxiv_org_abs_2309_10019
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts
Shi, Jiang-Xin
Wei, Tong
Zhou, Zhi
Shao, Jie-Jing
Han, Xin-Yan
Li, Yu-Feng
Computer Vision and Pattern Recognition
Machine Learning
The fine-tuning paradigm in addressing long-tail learning tasks has sparked significant interest since the emergence of foundation models. Nonetheless, how fine-tuning impacts performance in long-tail learning was not explicitly quantified. In this paper, we disclose that heavy fine-tuning may even lead to non-negligible performance deterioration on tail classes, and lightweight fine-tuning is more effective. The reason is attributed to inconsistent class conditions caused by heavy fine-tuning. With the observation above, we develop a low-complexity and accurate long-tail learning algorithms LIFT with the goal of facilitating fast prediction and compact models by adaptive lightweight fine-tuning. Experiments clearly verify that both the training time and the learned parameters are significantly reduced with more accurate predictive performance compared with state-of-the-art approaches. The implementation code is available at https://github.com/shijxcs/LIFT.
title Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2309.10019