Exploring Dualistic Meta-Learning to Enhance Domain Generalization in Open Set Scenarios

2026-06-22

AI Summary

This work addresses the challenge of decision boundary skew caused by mismatched label spaces between source and target domains in open-set domain generalization. To tackle this issue, the authors propose MEDIC, a dual meta-learning approach that, for the first time, jointly models domain alignment and category matching within a meta-learning framework. By leveraging implicit gradient matching, MEDIC simultaneously optimizes task partitions across domains and categories, while integrating a one-versus-rest classifier with an open-set anomaly detection mechanism to collaboratively enhance generalization in both open-set and closed-set scenarios. Experimental results demonstrate that MEDIC significantly outperforms existing methods on open-set domain generalization benchmarks and remains competitive under closed-set settings.

Abstract

Domain generalization learns from multiple source domains to generalize to unseen target domains. However, it often neglects the realistic case of label mismatch between source and target. Open set domain generalization is then proposed to recognize unseen classes in unseen domains. A simple approach trains one-vs-all classifiers to separate each class and detect outliers as unknown. Yet, the imbalance between few positive samples and many negative samples skews the decision boundary towards the positive ones, leading the model to over-reject out-of-distribution data, even from known classes in unseen domains. In this paper, we propose a novel meta-learning stategy called dualistic MEta-learning with joint DomaIn-Class matching (MEDIC), which considers implicit gradient matching towards inter-domain and inter-class task splits simultaneously to find optimal boundaries balanced for both domains and classes. Experimental results show that MEDIC not only outperforms prior methods in open set scenarios, but also maintains competitive close set generalization ability.

Problem

• domain generalization
• open set recognition
• label mismatch
• class imbalance
• out-of-distribution detection

Innovation

• meta-learning
• domain generalization
• open set recognition
• class imbalance
• gradient matching