Unsupervised Visual Domain Adaptation Using Subspace Alignment

TLDR

The paper proposes a new domain adaptation algorithm that represents source and target domains as subspaces defined by eigenvectors. The method learns a mapping that aligns the source subspace with the target, using a theoretical result to set the subspace size hyperparameter, yielding a closed‑form, extremely fast solution. Experiments on multiple datasets demonstrate that the closed‑form solution is extremely fast and outperforms state‑of‑the‑art domain adaptation methods.

Abstract

In this paper, we introduce a new domain adaptation (DA) algorithm where the source and target domains are represented by subspaces described by eigenvectors. In this context, our method seeks a domain adaptation solution by learning a mapping function which aligns the source subspace with the target one. We show that the solution of the corresponding optimization problem can be obtained in a simple closed form, leading to an extremely fast algorithm. We use a theoretical result to tune the unique hyper parameter corresponding to the size of the subspaces. We run our method on various datasets and show that, despite its intrinsic simplicity, it outperforms state of the art DA methods.