Residual Parameter Transfer for Deep Domain Adaptation

TL;DR

This paper introduces a residual parameter transfer method for deep domain adaptation, enabling flexible modeling of domain differences with improved accuracy and manageable complexity.

Contribution

It proposes a novel residual network architecture that predicts parameters for the target domain, enhancing adaptation flexibility over existing methods.

Findings

Achieves higher accuracy than state-of-the-art methods

Maintains manageable model complexity

Effectively models domain differences and similarities

Abstract

The goal of Deep Domain Adaptation is to make it possible to use Deep Nets trained in one domain where there is enough annotated training data in another where there is little or none. Most current approaches have focused on learning feature representations that are invariant to the changes that occur when going from one domain to the other, which means using the same network parameters in both domains. While some recent algorithms explicitly model the changes by adapting the network parameters, they either severely restrict the possible domain changes, or significantly increase the number of model parameters. By contrast, we introduce a network architecture that includes auxiliary residual networks, which we train to predict the parameters in the domain with little annotated data from those in the other one. This architecture enables us to flexibly preserve the similarities between domains where they exist and model the differences when necessary. We demonstrate that our approach yields higher accuracy than state-of-the-art methods without undue complexity.