Cross Domain Adaptation using Adversarial networks with Cyclic loss

TL;DR

This paper proposes a cyclic loss function for adversarial domain translation networks to improve their ability to adapt across different data domains, enhancing the accuracy of deep learning models in varied settings.

Contribution

Introduction of cyclic loss to constrain generator networks for more effective domain translation in adversarial settings.

Findings

Cyclic loss improves domain translation accuracy.

Enhanced generator stability with cyclic loss.

Better cross-domain generalization in deep learning models.

Abstract

Deep Learning methods are highly local and sensitive to the domain of data they are trained with. Even a slight deviation from the domain distribution affects prediction accuracy of deep networks significantly. In this work, we have investigated a set of techniques aimed at increasing accuracy of generator networks which perform translation from one domain to the other in an adversarial setting. In particular, we experimented with activations, the encoder-decoder network architectures, and introduced a Loss called cyclic loss to constrain the Generator network so that it learns effective source-target translation. This machine learning problem is motivated by myriad applications that can be derived from domain adaptation networks like generating labeled data from synthetic inputs in an unsupervised fashion, and using these translation network in conjunction with the original domain network to generalize deep learning networks across domains.