Unified Deep Supervised Domain Adaptation and Generalization

TLDR

The supervised setting is attractive when only a few target samples are labeled, yet aligning and separating semantic probability distributions is difficult due to data scarcity. This work proposes a unified framework for visual supervised domain adaptation and generalization using deep models. The method uses a Siamese architecture to learn a discriminative embedding subspace that aligns mapped visual domains while keeping them maximally separated, and it is extended to domain generalization. Experiments show that using point‑wise surrogates of distribution distances and similarities yields an effective solution, the approach adapts quickly with as few as one labeled target sample per category, and overall results are very promising for both domain adaptation and generalization.

Abstract

This work provides a unified framework for addressing the problem of visual supervised domain adaptation and generalization with deep models. The main idea is to exploit the Siamese architecture to learn an embedding subspace that is discriminative, and where mapped visual domains are semantically aligned and yet maximally separated. The supervised setting becomes attractive especially when only few target data samples need to be labeled. In this scenario, alignment and separation of semantic probability distributions is difficult because of the lack of data. We found that by reverting to point-wise surrogates of distribution distances and similarities provides an effective solution. In addition, the approach has a high "speed" of adaptation, which requires an extremely low number of labeled target training samples, even one per category can be effective. The approach is extended to domain generalization. For both applications the experiments show very promising results.