Adaptive Prompt Learning with Negative Textual Semantics and Uncertainty Modeling for Universal Multi-Source Domain Adaptation

TL;DR

This paper introduces APNE-CLIP, a novel method for universal multi-source domain adaptation that leverages adaptive prompts, negative textual semantics, and uncertainty modeling to improve unknown sample detection and domain shift handling.

Contribution

It proposes a new approach combining adaptive prompt learning, negative textual semantics, and energy-based uncertainty modeling within CLIP for improved UniMDA performance.

Findings

Outperforms existing methods in UniMDA tasks

Effectively detects unknown samples using negative textual semantics

Enhances domain adaptation with uncertainty modeling

Abstract

Universal Multi-source Domain Adaptation (UniMDA) transfers knowledge from multiple labeled source domains to an unlabeled target domain under domain shifts (different data distribution) and class shifts (unknown target classes). Existing solutions focus on excavating image features to detect unknown samples, ignoring abundant information contained in textual semantics. In this paper, we propose an Adaptive Prompt learning with Negative textual semantics and uncErtainty modeling method based on Contrastive Language-Image Pre-training (APNE-CLIP) for UniMDA classification tasks. Concretely, we utilize the CLIP with adaptive prompts to leverage textual information of class semantics and domain representations, helping the model identify unknown samples and address domain shifts. Additionally, we design a novel global instance-level alignment objective by utilizing negative textual semantics to achieve more precise image-text pair alignment. Furthermore, we propose an energy-based uncertainty modeling strategy to enlarge the margin distance between known and unknown samples. Extensive experiments demonstrate the superiority of our proposed method.