Deep Domain Adaptation by Geodesic Distance Minimization
This is an incremental improvement for domain adaptation in computer vision, addressing the problem of adapting models across different visual domains.
The paper tackles unsupervised visual domain adaptation by proposing Deep LogCORAL, which replaces Euclidean distance with Riemannian distance and includes first-order information to minimize domain differences, resulting in improved performance over Deep CORAL on the Office dataset.
In this paper, we propose a new approach called Deep LogCORAL for unsupervised visual domain adaptation. Our work builds on the recently proposed Deep CORAL method, which proposed to train a convolutional neural network and simultaneously minimize the Euclidean distance of convariance matrices between the source and target domains. We propose to use the Riemannian distance, approximated by Log-Euclidean distance, to replace the naive Euclidean distance in Deep CORAL. We also consider first-order information, and minimize the distance of mean vectors between two domains. We build an end-to-end model, in which we minimize both the classification loss, and the domain difference based on the first and second order information between two domains. Our experiments on the benchmark Office dataset demonstrate the improvements of our newly proposed Deep LogCORAL approach over the Deep CORAL method, as well as further improvement when optimizing both orders of information.