Semantic Bridging Domains: Pseudo-Source as Test-Time Connector

2026-03-04

AI Summary

This work addresses the challenging problem of test-time domain shift where the source domain is unknown and the target domain lacks labeled data. The authors propose Stepwise Semantic Alignment (SSA), a novel approach that treats a pseudo-source domain not as a direct substitute for the original source but as a semantic intermediary bridging the source and target domains. SSA employs a general semantics-guided, stepwise alignment mechanism to first rectify features from the pseudo-source domain before aligning them with the target domain. To further enhance alignment quality, the method integrates Hierarchical Feature Aggregation (HFA) and Confidence-Aware Complementary Learning (CACL). Extensive experiments demonstrate significant performance gains: SSA outperforms the current state-of-the-art by 5.2% on the GTA→Cityscapes semantic segmentation benchmark and achieves notable improvements in image classification tasks under domain shift.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Segmentation
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance

Abstract

Distribution shifts between training and testing data are a critical bottleneck limiting the practical utility of models, especially in real-world test-time scenarios. To adapt models when the source domain is unknown and the target domain is unlabeled, previous works constructed pseudo-source domains via data generation and translation, then aligned the target domain with them. However, significant discrepancies exist between the pseudo-source and the original source domain, leading to potential divergence when correcting the target directly. From this perspective, we propose a Stepwise Semantic Alignment (SSA) method, viewing the pseudo-source as a semantic bridge connecting the source and target, rather than a direct substitute for the source. Specifically, we leverage easily accessible universal semantics to rectify the semantic features of the pseudo-source, and then align the target domain using the corrected pseudo-source semantics. Additionally, we introduce a Hierarchical Feature Aggregation (HFA) module and a Confidence-Aware Complementary Learning (CACL) strategy to enhance the semantic quality of the SSA process in the absence of source and ground truth of target domains. We evaluated our approach on tasks like semantic segmentation and image classification, achieving a 5.2% performance boost on GTA2Cityscapes over the state-of-the-art.

Problem

• distribution shift
• domain adaptation
• pseudo-source domain
• unlabeled target domain
• semantic alignment

Innovation

• Stepwise Semantic Alignment
• Pseudo-Source Domain
• Hierarchical Feature Aggregation
• Confidence-Aware Complementary Learning
• Test-Time Adaptation