Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yang, Yiting, Luo, Hao, Sun, Yuan, Yan, Qingsen, Zhang, Haokui, Dong, Wei, Wang, Guoqing, Wang, Peng, Yang, Yang, Shen, Hengtao
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908454962069504
author Yang, Yiting
Luo, Hao
Sun, Yuan
Yan, Qingsen
Zhang, Haokui
Dong, Wei
Wang, Guoqing
Wang, Peng
Yang, Yang
Shen, Hengtao
author_facet Yang, Yiting
Luo, Hao
Sun, Yuan
Yan, Qingsen
Zhang, Haokui
Dong, Wei
Wang, Guoqing
Wang, Peng
Yang, Yang
Shen, Hengtao
contents A prevalent approach in Parameter-Efficient Fine-Tuning (PEFT) of pre-trained Vision Transformers (ViT) involves freezing the majority of the backbone parameters and solely learning low-rank adaptation weight matrices to accommodate downstream tasks. These low-rank matrices are commonly derived through the multiplication structure of down-projection and up-projection matrices, exemplified by methods such as LoRA and Adapter. In this work, we observe an approximate orthogonality among any two row or column vectors within any weight matrix of the backbone parameters; however, this property is absent in the vectors of the down/up-projection matrices. Approximate orthogonality implies a reduction in the upper bound of the model's generalization error, signifying that the model possesses enhanced generalization capability. If the fine-tuned down/up-projection matrices were to exhibit this same property as the pre-trained backbone matrices, could the generalization capability of fine-tuned ViTs be further augmented? To address this question, we propose an Approximately Orthogonal Fine-Tuning (AOFT) strategy for representing the low-rank weight matrices. This strategy employs a single learnable vector to generate a set of approximately orthogonal vectors, which form the down/up-projection matrices, thereby aligning the properties of these matrices with those of the backbone. Extensive experimental results demonstrate that our method achieves competitive performance across a range of downstream image classification tasks, confirming the efficacy of the enhanced generalization capability embedded in the down/up-projection matrices.
format Preprint
id arxiv_https___arxiv_org_abs_2507_13260
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy
Yang, Yiting
Luo, Hao
Sun, Yuan
Yan, Qingsen
Zhang, Haokui
Dong, Wei
Wang, Guoqing
Wang, Peng
Yang, Yang
Shen, Hengtao
Computer Vision and Pattern Recognition
Artificial Intelligence
A prevalent approach in Parameter-Efficient Fine-Tuning (PEFT) of pre-trained Vision Transformers (ViT) involves freezing the majority of the backbone parameters and solely learning low-rank adaptation weight matrices to accommodate downstream tasks. These low-rank matrices are commonly derived through the multiplication structure of down-projection and up-projection matrices, exemplified by methods such as LoRA and Adapter. In this work, we observe an approximate orthogonality among any two row or column vectors within any weight matrix of the backbone parameters; however, this property is absent in the vectors of the down/up-projection matrices. Approximate orthogonality implies a reduction in the upper bound of the model's generalization error, signifying that the model possesses enhanced generalization capability. If the fine-tuned down/up-projection matrices were to exhibit this same property as the pre-trained backbone matrices, could the generalization capability of fine-tuned ViTs be further augmented? To address this question, we propose an Approximately Orthogonal Fine-Tuning (AOFT) strategy for representing the low-rank weight matrices. This strategy employs a single learnable vector to generate a set of approximately orthogonal vectors, which form the down/up-projection matrices, thereby aligning the properties of these matrices with those of the backbone. Extensive experimental results demonstrate that our method achieves competitive performance across a range of downstream image classification tasks, confirming the efficacy of the enhanced generalization capability embedded in the down/up-projection matrices.
title Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2507.13260