SfMamba: Efficient Source-Free Domain Adaptation via Selective Scan Modeling

2026-01-13

AI Summary

This work proposes an efficient and parameter-light source-free domain adaptation method under the challenging setting where source-domain data are unavailable. By introducing a channel-wise visual state space block to model channel-frequency characteristics, the approach integrates Mamba-based selective scanning with a semantic-consistent shuffling strategy to enhance feature alignment and spatial robustness without accessing source data, thereby effectively mitigating error accumulation. The method achieves significant performance gains over existing techniques across multiple benchmarks, demonstrating a favorable balance between computational efficiency and adaptation accuracy. This study thus offers a practical and scalable solution to source-free domain adaptation, advancing its applicability in real-world scenarios where source data privacy or storage constraints preclude their use.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Diffusion Models for Vision
• 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
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Source-free domain adaptation (SFDA) tackles the critical challenge of adapting source-pretrained models to unlabeled target domains without access to source data, overcoming data privacy and storage limitations in real-world applications. However, existing SFDA approaches struggle with the trade-off between perception field and computational efficiency in domain-invariant feature learning. Recently, Mamba has offered a promising solution through its selective scan mechanism, which enables long-range dependency modeling with linear complexity. However, the Visual Mamba (i.e., VMamba) remains limited in capturing channel-wise frequency characteristics critical for domain alignment and maintaining spatial robustness under significant domain shifts. To address these, we propose a framework called SfMamba to fully explore the stable dependency in source-free model transfer. SfMamba introduces Channel-wise Visual State-Space block that enables channel-sequence scanning for domain-invariant feature extraction. In addition, SfMamba involves a Semantic-Consistent Shuffle strategy that disrupts background patch sequences in 2D selective scan while preserving prediction consistency to mitigate error accumulation. Comprehensive evaluations across multiple benchmarks show that SfMamba achieves consistently stronger performance than existing methods while maintaining favorable parameter efficiency, offering a practical solution for SFDA. Our code is available at https://github.com/chenxi52/SfMamba.

Problem

• Source-Free Domain Adaptation
• Domain-Invariant Feature Learning
• Selective Scan
• Channel-wise Frequency Characteristics
• Spatial Robustness

Innovation

• Source-Free Domain Adaptation
• Selective Scan
• Channel-wise Modeling
• Semantic-Consistent Shuffle
• Visual Mamba