Multi-Source Domain Adaptation through Dataset Dictionary Learning in Wasserstein Space
It addresses data distribution shifts for transferring knowledge from multiple labeled source domains to an unlabeled target domain, with incremental improvements in domain adaptation.
This paper tackles Multi-Source Domain Adaptation (MSDA) by proposing a framework based on dictionary learning and optimal transport, interpreting domains as empirical distributions and expressing them as Wasserstein barycenters. The methods DaDiL-R and DaDiL-E improved state-of-the-art classification performance by 3.15%, 2.29%, and 7.71% on three benchmarks.
This paper seeks to solve Multi-Source Domain Adaptation (MSDA), which aims to mitigate data distribution shifts when transferring knowledge from multiple labeled source domains to an unlabeled target domain. We propose a novel MSDA framework based on dictionary learning and optimal transport. We interpret each domain in MSDA as an empirical distribution. As such, we express each domain as a Wasserstein barycenter of dictionary atoms, which are empirical distributions. We propose a novel algorithm, DaDiL, for learning via mini-batches: (i) atom distributions; (ii) a matrix of barycentric coordinates. Based on our dictionary, we propose two novel methods for MSDA: DaDil-R, based on the reconstruction of labeled samples in the target domain, and DaDiL-E, based on the ensembling of classifiers learned on atom distributions. We evaluate our methods in 3 benchmarks: Caltech-Office, Office 31, and CRWU, where we improved previous state-of-the-art by 3.15%, 2.29%, and 7.71% in classification performance. Finally, we show that interpolations in the Wasserstein hull of learned atoms provide data that can generalize to the target domain.