Increasing Model Generalizability for Unsupervised Visual Domain Adaptation

Mohammad Rostami
Proceedings of The 1st Conference on Lifelong Learning Agents, PMLR 199:281-293, 2022.

Abstract

A dominant approach for addressing unsupervised domain adaptation is to map data points for the source and the target domains into an embedding space which is modeled as the output-space of a shared deep encoder. The encoder is trained to make the embedding space domain-agnostic to make a source-trained classifier generalizable on the target domain. A secondary mechanism to improve UDA performance further is to make the source domain distribution more compact to improve model generalizability. We demonstrate that increasing the interclass margins in the embedding space can help to develop a UDA algorithm with improved performance. We estimate the internally learned multi-modal distribution for the source domain, learned as a result of pretraining, and use it to increase the interclass class separation in the source domain to reduce the effect of domain shift. We demonstrate that using our approach leads to improved model generalizability on four standard benchmark UDA image classification datasets and compares favorably against exiting methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v199-rostami22a, title = {Increasing Model Generalizability for Unsupervised Visual Domain Adaptation}, author = {Rostami, Mohammad}, booktitle = {Proceedings of The 1st Conference on Lifelong Learning Agents}, pages = {281--293}, year = {2022}, editor = {Chandar, Sarath and Pascanu, Razvan and Precup, Doina}, volume = {199}, series = {Proceedings of Machine Learning Research}, month = {22--24 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v199/rostami22a/rostami22a.pdf}, url = {https://proceedings.mlr.press/v199/rostami22a.html}, abstract = {A dominant approach for addressing unsupervised domain adaptation is to map data points for the source and the target domains into an embedding space which is modeled as the output-space of a shared deep encoder. The encoder is trained to make the embedding space domain-agnostic to make a source-trained classifier generalizable on the target domain. A secondary mechanism to improve UDA performance further is to make the source domain distribution more compact to improve model generalizability. We demonstrate that increasing the interclass margins in the embedding space can help to develop a UDA algorithm with improved performance. We estimate the internally learned multi-modal distribution for the source domain, learned as a result of pretraining, and use it to increase the interclass class separation in the source domain to reduce the effect of domain shift. We demonstrate that using our approach leads to improved model generalizability on four standard benchmark UDA image classification datasets and compares favorably against exiting methods.} }
Endnote
%0 Conference Paper %T Increasing Model Generalizability for Unsupervised Visual Domain Adaptation %A Mohammad Rostami %B Proceedings of The 1st Conference on Lifelong Learning Agents %C Proceedings of Machine Learning Research %D 2022 %E Sarath Chandar %E Razvan Pascanu %E Doina Precup %F pmlr-v199-rostami22a %I PMLR %P 281--293 %U https://proceedings.mlr.press/v199/rostami22a.html %V 199 %X A dominant approach for addressing unsupervised domain adaptation is to map data points for the source and the target domains into an embedding space which is modeled as the output-space of a shared deep encoder. The encoder is trained to make the embedding space domain-agnostic to make a source-trained classifier generalizable on the target domain. A secondary mechanism to improve UDA performance further is to make the source domain distribution more compact to improve model generalizability. We demonstrate that increasing the interclass margins in the embedding space can help to develop a UDA algorithm with improved performance. We estimate the internally learned multi-modal distribution for the source domain, learned as a result of pretraining, and use it to increase the interclass class separation in the source domain to reduce the effect of domain shift. We demonstrate that using our approach leads to improved model generalizability on four standard benchmark UDA image classification datasets and compares favorably against exiting methods.
APA
Rostami, M.. (2022). Increasing Model Generalizability for Unsupervised Visual Domain Adaptation. Proceedings of The 1st Conference on Lifelong Learning Agents, in Proceedings of Machine Learning Research 199:281-293 Available from https://proceedings.mlr.press/v199/rostami22a.html.

Related Material