BaFTA: Backprop-Free Test-Time Adaptation For Zero-Shot Vision-Language Models

TL;DR

BaFTA is a novel backpropagation-free method for test-time adaptation of vision-language models like CLIP, using online clustering and entropy-based reliability to improve zero-shot image classification without fine-tuning.

Contribution

It introduces a backpropagation-free algorithm that estimates class centroids via online clustering, avoiding fine-tuning and improving adaptation performance.

Findings

BaFTA outperforms existing methods in accuracy.

BaFTA is more efficient due to no backpropagation.

It effectively combines multiple predictions for robustness.

Abstract

Large-scale pretrained vision-language models like CLIP have demonstrated remarkable zero-shot image classification capabilities across diverse domains. To enhance CLIP's performance while preserving the zero-shot paradigm, various test-time prompt tuning methods have been introduced to refine class embeddings through unsupervised learning objectives during inference. However, these methods often encounter challenges in selecting appropriate learning rates to prevent collapsed training in the absence of validation data during test-time adaptation. In this study, we propose a novel backpropagation-free algorithm BaFTA for test-time adaptation of vision-language models. Instead of fine-tuning text prompts to refine class embeddings, our approach directly estimates class centroids using online clustering within a projected embedding space that aligns text and visual embeddings. We dynamically aggregate predictions from both estimated and original class embeddings, as well as from distinct augmented views, by assessing the reliability of each prediction using R\'enyi Entropy. Through extensive experiments, we demonstrate that BaFTA consistently outperforms state-of-the-art test-time adaptation methods in both effectiveness and efficiency.