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

The paper introduces a novel meta-learning strategy called MEDIC (dualistic MEta-learning with joint DomaIn-Class matching) aimed at enhancing domain generalization in open set scenarios, where label mismatches occur between source and target domains. MEDIC employs implicit gradient matching to optimize decision boundaries for both domains and classes, addressing the imbalance in sample distribution that affects traditional one-vs-all classifiers. Experimental results demonstrate that MEDIC outperforms existing methods in open set scenarios while retaining competitive performance in closed set generalization, making it a valuable approach for practitioners dealing with unseen classes in real-world applications.