ProKeR: A Kernel Perspective on Few-Shot Adaptation of Large Vision-Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bendou, Yassir, Ouasfi, Amine, Gripon, Vincent, Boukhayma, Adnane
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929682258067456
author Bendou, Yassir
Ouasfi, Amine
Gripon, Vincent
Boukhayma, Adnane
author_facet Bendou, Yassir
Ouasfi, Amine
Gripon, Vincent
Boukhayma, Adnane
contents The growing popularity of Contrastive Language-Image Pretraining (CLIP) has led to its widespread application in various visual downstream tasks. To enhance CLIP's effectiveness and versatility, efficient few-shot adaptation techniques have been widely adopted. Among these approaches, training-free methods, particularly caching methods exemplified by Tip-Adapter, have gained attention for their lightweight adaptation without the need for additional fine-tuning. In this paper, we revisit Tip-Adapter from a kernel perspective, showing that caching methods function as local adapters and are connected to a well-established kernel literature. Drawing on this insight, we offer a theoretical understanding of how these methods operate and suggest multiple avenues for enhancing the Tip-Adapter baseline. Notably, our analysis shows the importance of incorporating global information in local adapters. Therefore, we subsequently propose a global method that learns a proximal regularizer in a reproducing kernel Hilbert space (RKHS) using CLIP as a base learner. Our method, which we call ProKeR (Proximal Kernel ridge Regression), has a closed form solution and achieves state-of-the-art performances across 11 datasets in the standard few-shot adaptation benchmark.
format Preprint
id arxiv_https___arxiv_org_abs_2501_11175
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ProKeR: A Kernel Perspective on Few-Shot Adaptation of Large Vision-Language Models
Bendou, Yassir
Ouasfi, Amine
Gripon, Vincent
Boukhayma, Adnane
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
The growing popularity of Contrastive Language-Image Pretraining (CLIP) has led to its widespread application in various visual downstream tasks. To enhance CLIP's effectiveness and versatility, efficient few-shot adaptation techniques have been widely adopted. Among these approaches, training-free methods, particularly caching methods exemplified by Tip-Adapter, have gained attention for their lightweight adaptation without the need for additional fine-tuning. In this paper, we revisit Tip-Adapter from a kernel perspective, showing that caching methods function as local adapters and are connected to a well-established kernel literature. Drawing on this insight, we offer a theoretical understanding of how these methods operate and suggest multiple avenues for enhancing the Tip-Adapter baseline. Notably, our analysis shows the importance of incorporating global information in local adapters. Therefore, we subsequently propose a global method that learns a proximal regularizer in a reproducing kernel Hilbert space (RKHS) using CLIP as a base learner. Our method, which we call ProKeR (Proximal Kernel ridge Regression), has a closed form solution and achieves state-of-the-art performances across 11 datasets in the standard few-shot adaptation benchmark.
title ProKeR: A Kernel Perspective on Few-Shot Adaptation of Large Vision-Language Models
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2501.11175