Patch-Prompt Aligned Bayesian Prompt Tuning for Vision-Language Models

Xinyang Liu, Dongsheng Wang, Bowei Fang, Miaoge Li, Yishi Xu, Zhibin Duan, Bo Chen, Mingyuan Zhou

Keywords: Bayesian prompt learning; pre-trained vision language model

TL;DR: We introduce a Bayesian prompt learning that learns class-specific stochastic prompts for CLIP.

Abstract: For downstream applications of vision-language pre-trained models, there has been significant interest in constructing effective prompts. Existing works on prompt engineering, which either require laborious manual designs or optimize the prompt tuning as a point estimation problem, may fail to describe diverse characteristics of categories and limit their applications. We introduce a Bayesian probabilistic resolution to prompt tuning, where the label-specific stochastic prompts are generated hierarchically by first sampling a latent vector from an underlying distribution and then employing a lightweight generative model. Importantly, we semantically regularize the tuning process by minimizing the statistic distance between the visual patches and linguistic prompts, which pushes the stochastic label representations to faithfully capture diverse visual concepts, instead of overfitting the training categories. We evaluate the effectiveness of our approach on four tasks: few-shot image recognition, base-to-new generalization, dataset transfer learning, and domain shifts. Extensive results on over 15 datasets show promising transferability and generalization performance of our proposed model, both quantitatively and qualitatively.

List Of Authors: Liu, Xinyang and Wang, Dongsheng and Fang, Bowei and Li, Miaoge and Xu, Yishi and Duan, Zhibin and Chen, Bo and Zhou, Mingyuan

Submission Number: 223