Domain Generalization via Invariant Feature Representation
This addresses the problem of applying learned knowledge to new, unseen domains for machine learning practitioners, representing an incremental advance in domain generalization methods.
The paper tackles domain generalization by proposing Domain-Invariant Component Analysis (DICA), a kernel-based method that learns invariant transformations to improve classifier performance on unseen domains, with experimental validation on synthetic and real-world datasets.
This paper investigates domain generalization: How to take knowledge acquired from an arbitrary number of related domains and apply it to previously unseen domains? We propose Domain-Invariant Component Analysis (DICA), a kernel-based optimization algorithm that learns an invariant transformation by minimizing the dissimilarity across domains, whilst preserving the functional relationship between input and output variables. A learning-theoretic analysis shows that reducing dissimilarity improves the expected generalization ability of classifiers on new domains, motivating the proposed algorithm. Experimental results on synthetic and real-world datasets demonstrate that DICA successfully learns invariant features and improves classifier performance in practice.