Supervised Domain Adaptation: A Graph Embedding Perspective and a Rectified Experimental Protocol

The performance of machine learning models tends to suffer when the distributions of the training and test data differ. Domain Adaptation is the process of closing the distribution gap between datasets. In this paper, we show that Domain Adaptation methods using pair-wise relationships between source and target domain data can be formulated as a Graph Embedding in which the domain labels are incorporated into the structure of the intrinsic and penalty graphs. We analyse the loss functions of existing state-of-the-art Supervised Domain Adaptation methods and demonstrate that they perform Graph Embedding. Moreover, we highlight some generalisation and reproducibility issues related to the experimental setup commonly used to demonstrate the few-shot learning capabilities of these methods. We propose a rectified evaluation setup for more accurately assessing and comparing Supervised Domain Adaptation methods, and report experiments on the standard benchmark datasets Office31 and MNIST-USPS.