Related papers: Multi-Prompt Progressive Alignment for Multi-Source Unsupervised Domain Adaptation

Multi-Prompt Progressive Alignment for Multi-Source Unsupervised Domain Adaptation

URL: http://arxiv.org/abs/2507.23373v1
Date: Thu, 31 Jul 2025 09:42:42 GMT
Title: Multi-Prompt Progressive Alignment for Multi-Source Unsupervised Domain Adaptation
Authors: Haoran Chen, Zexiao Wang, Haidong Cao, Zuxuan Wu, Yu-Gang Jiang,
Abstract summary: We propose a progressive alignment strategy for adapting CLIP to unlabeled downstream task.<n>We name our approach MP2A and test it on three popular UDA benchmarks, namely ImageCLEF, Office-Home, and the most challenging DomainNet.<n> Experiments showcase that MP2A achieves state-of-the-art performance when compared with most recent CLIP-based MS-UDA approaches.
Score: 73.40696661117408
License: http://creativecommons.org/licenses/by/4.0/
Abstract: Large Vision-Language Models like CLIP have become a powerful foundation for Unsupervised Domain Adaptation due to their strong zero-shot generalization. State-of-the-art methods typically leverage CLIP to generate pseudo-labels for the target domain, then fine-tune the model to learn domain-invariant features. However, these methods attempt to align source and target domains using all pseudo-labeled data simultaneously. This one-shot alignment struggles with noisy, hard-to-classify samples, leading to error propagation and suboptimal feature learning. The problem is even more amplified in the multi-source scenario, where diverse domain gaps and varying noise levels across multiple source domains further destabilize the alignment process. To address this issue, in this work, we propose a progressive alignment strategy for adapting CLIP to unlabeled downstream task. Our method begins by training the model on a high-confidence subset of target samples, allowing it to first learn a well-aligned representation from the most reliable data. As training progresses, it gradually incorporates more challenging samples, guiding the model to refine its understanding without being overwhelmed by initial label noise. This progressive approach effectively mitigates confirmation bias and promotes a more robust convergence, allowing for the learning of genuinely domain-invariant features. We name our approach MP^2A and test it on three popular UDA benchmarks, namely ImageCLEF, Office-Home, and the most challenging DomainNet. Experiments showcase that MP^2A achieves state-of-the-art performance when compared with most recent CLIP-based MS-UDA approaches, demonstrating the effectiveness of our approach.

Related papers

Improving Domain Adaptation Through Class Aware Frequency Transformation [15.70058524548143]
Most of the Unsupervised Domain Adaptation (UDA) algorithms focus on reducing the global domain shift between labelled source and unlabelled target domains. We propose a novel approach based on traditional image processing technique Class Aware Frequency Transformation (CAFT) CAFT utilizes pseudo label based class consistent low-frequency swapping for improving the overall performance of the existing UDA algorithms.
arXiv Detail & Related papers (2024-07-28T18:16:41Z)
StyDeSty: Min-Max Stylization and Destylization for Single Domain Generalization [85.18995948334592]
Single domain generalization (single DG) aims at learning a robust model generalizable to unseen domains from only one training domain. State-of-the-art approaches have mostly relied on data augmentations, such as adversarial perturbation and style enhancement, to synthesize new data. We propose emphStyDeSty, which explicitly accounts for the alignment of the source and pseudo domains in the process of data augmentation.
arXiv Detail & Related papers (2024-06-01T02:41:34Z)
Divide and Contrast: Source-free Domain Adaptation via Adaptive Contrastive Learning [122.62311703151215]
Divide and Contrast (DaC) aims to connect the good ends of both worlds while bypassing their limitations. DaC divides the target data into source-like and target-specific samples, where either group of samples is treated with tailored goals. We further align the source-like domain with the target-specific samples using a memory bank-based Maximum Mean Discrepancy (MMD) loss to reduce the distribution mismatch.
arXiv Detail & Related papers (2022-11-12T09:21:49Z)
Low-confidence Samples Matter for Domain Adaptation [47.552605279925736]
Domain adaptation (DA) aims to transfer knowledge from a label-rich source domain to a related but label-scarce target domain. We propose a novel contrastive learning method by processing low-confidence samples. We evaluate the proposed method in both unsupervised and semi-supervised DA settings.
arXiv Detail & Related papers (2022-02-06T15:45:45Z)
Semi-Supervised Domain Adaptation with Prototypical Alignment and Consistency Learning [86.6929930921905]
This paper studies how much it can help address domain shifts if we further have a few target samples labeled. To explore the full potential of landmarks, we incorporate a prototypical alignment (PA) module which calculates a target prototype for each class from the landmarks. Specifically, we severely perturb the labeled images, making PA non-trivial to achieve and thus promoting model generalizability.
arXiv Detail & Related papers (2021-04-19T08:46:08Z)
Instance Level Affinity-Based Transfer for Unsupervised Domain Adaptation [74.71931918541748]
We propose an instance affinity based criterion for source to target transfer during adaptation, called ILA-DA. We first propose a reliable and efficient method to extract similar and dissimilar samples across source and target, and utilize a multi-sample contrastive loss to drive the domain alignment process. We verify the effectiveness of ILA-DA by observing consistent improvements in accuracy over popular domain adaptation approaches on a variety of benchmark datasets.
arXiv Detail & Related papers (2021-04-03T01:33:14Z)
Adaptive Pseudo-Label Refinement by Negative Ensemble Learning for Source-Free Unsupervised Domain Adaptation [35.728603077621564]
Existing Unsupervised Domain Adaptation (UDA) methods presumes source and target domain data to be simultaneously available during training. A pre-trained source model is always considered to be available, even though performing poorly on target due to the well-known domain shift problem. We propose a unified method to tackle adaptive noise filtering and pseudo-label refinement.
arXiv Detail & Related papers (2021-03-29T22:18:34Z)
Surprisingly Simple Semi-Supervised Domain Adaptation with Pretraining and Consistency [93.89773386634717]
Visual domain adaptation involves learning to classify images from a target visual domain using labels available in a different source domain. We show that in the presence of a few target labels, simple techniques like self-supervision (via rotation prediction) and consistency regularization can be effective without any adversarial alignment to learn a good target classifier. Our Pretraining and Consistency (PAC) approach, can achieve state of the art accuracy on this semi-supervised domain adaptation task, surpassing multiple adversarial domain alignment methods, across multiple datasets.
arXiv Detail & Related papers (2021-01-29T18:40:17Z)
Discrepancy Minimization in Domain Generalization with Generative Nearest Neighbors [13.047289562445242]
Domain generalization (DG) deals with the problem of domain shift where a machine learning model trained on multiple-source domains fail to generalize well on a target domain with different statistics. Multiple approaches have been proposed to solve the problem of domain generalization by learning domain invariant representations across the source domains that fail to guarantee generalization on the shifted target domain. We propose a Generative Nearest Neighbor based Discrepancy Minimization (GNNDM) method which provides a theoretical guarantee that is upper bounded by the error in the labeling process of the target.
arXiv Detail & Related papers (2020-07-28T14:54:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.