E-MLNet: Enhanced Mutual Learning for Universal Domain Adaptation with Sample-Specific Weighting

2025-09-10

AI Summary

Universal Domain Adaptation (UniDA) requires models to jointly perform known-class classification and unknown-class rejection under settings where source-domain labels are available, target-domain labels are absent, and the target-class set is entirely unknown. Existing mutual learning approaches (e.g., MLNet) suffer from diluted learning signals due to uniform weighting across all classifiers. To address this, we propose a dynamic weighted mutual learning framework: (1) a one-vs-all classifier bank is constructed, with sample-level prediction confidence used to adaptively focus on the most relevant class boundaries, thereby enhancing open-set entropy minimization; and (2) a sample-specific weighting mechanism is introduced to improve discriminative boundary modeling. Evaluated on VisDA and ImageCLEF benchmarks under both open-partial and open-set DA settings, our method outperforms MLNet on 22/31 and 19/31 tasks, respectively, achieving state-of-the-art average H-score.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Universal Domain Adaptation (UniDA) seeks to transfer knowledge from a labeled source to an unlabeled target domain without assuming any relationship between their label sets, requiring models to classify known samples while rejecting unknown ones. Advanced methods like Mutual Learning Network (MLNet) use a bank of one-vs-all classifiers adapted via Open-set Entropy Minimization (OEM). However, this strategy treats all classifiers equally, diluting the learning signal. We propose the Enhanced Mutual Learning Network (E-MLNet), which integrates a dynamic weighting strategy to OEM. By leveraging the closed-set classifier's predictions, E-MLNet focuses adaptation on the most relevant class boundaries for each target sample, sharpening the distinction between known and unknown classes. We conduct extensive experiments on four challenging benchmarks: Office-31, Office-Home, VisDA-2017, and ImageCLEF. The results demonstrate that E-MLNet achieves the highest average H-scores on VisDA and ImageCLEF and exhibits superior robustness over its predecessor. E-MLNet outperforms the strong MLNet baseline in the majority of individual adaptation tasks -- 22 out of 31 in the challenging Open-Partial DA setting and 19 out of 31 in the Open-Set DA setting -- confirming the benefits of our focused adaptation strategy.

Problem

• Enhancing mutual learning for universal domain adaptation
• Addressing equal classifier treatment in existing methods
• Improving known-unknown class distinction with dynamic weighting

Innovation

• Dynamic weighting strategy for OEM
• Sample-specific class boundary adaptation
• Enhanced mutual learning network