Sliced Wasserstein Discrepancy in Disentangling Representation and Adaptation Networks for Unsupervised Domain Adaptation
Joel Sol, Shadi Alijani, Homayoun Najjaran
TL;DR
Unsupervised domain adaptation struggles with cross-domain shifts between labeled sources and unlabeled targets. The authors propose DRANet-SWD by substituting the Gram matrix style loss with the sliced Wasserstein discrepancy to better capture full feature distributions during content style disentanglement. They validate on digit recognition tasks across MNIST, MNIST-M, USPS, and SVHN and on driving-scene segmentation from GTA5 to Cityscapes, showing improved alignment and performance in most settings. The work highlights the value of distribution-aware style losses for robust domain adaptation and sets up future directions in representation learning and hyperparameter tuning for harder transfers.
Abstract
This paper introduces DRANet-SWD as a novel complete pipeline for disentangling content and style representations of images for unsupervised domain adaptation (UDA). The approach builds upon DRANet by incorporating the sliced Wasserstein discrepancy (SWD) as a style loss instead of the traditional Gram matrix loss. The potential advantages of SWD over the Gram matrix loss for capturing style variations in domain adaptation are investigated. Experiments using digit classification datasets and driving scenario segmentation validate the method, demonstrating that DRANet-SWD enhances performance. Results indicate that SWD provides a more robust statistical comparison of feature distributions, leading to better style adaptation. These findings highlight the effectiveness of SWD in refining feature alignment and improving domain adaptation tasks across these benchmarks. Our code can be found here.