Domain Aggregation Networks for Multi-Source Domain Adaptation

Junfeng Wen, Russell Greiner, Dale Schuurmans
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:10214-10224, 2020.

Abstract

In many real-world applications, we want to exploit multiple source datasets to build a model for a different but related target dataset. Despite the recent empirical success, most existing research has used ad-hoc methods to combine multiple sources, leading to a gap between theory and practice. In this paper, we develop a finite-sample generalization bound based on domain discrepancy and accordingly propose a theoretically justified optimization procedure. Our algorithm, Domain AggRegation Network (DARN), can automatically and dynamically balance between including more data to increase effective sample size and excluding irrelevant data to avoid negative effects during training. We find that DARN can significantly outperform the state-of-the-art alternatives on multiple real-world tasks, including digit/object recognition and sentiment analysis.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-wen20b, title = {Domain Aggregation Networks for Multi-Source Domain Adaptation}, author = {Wen, Junfeng and Greiner, Russell and Schuurmans, Dale}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {10214--10224}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/wen20b/wen20b.pdf}, url = {http://proceedings.mlr.press/v119/wen20b.html}, abstract = {In many real-world applications, we want to exploit multiple source datasets to build a model for a different but related target dataset. Despite the recent empirical success, most existing research has used ad-hoc methods to combine multiple sources, leading to a gap between theory and practice. In this paper, we develop a finite-sample generalization bound based on domain discrepancy and accordingly propose a theoretically justified optimization procedure. Our algorithm, Domain AggRegation Network (DARN), can automatically and dynamically balance between including more data to increase effective sample size and excluding irrelevant data to avoid negative effects during training. We find that DARN can significantly outperform the state-of-the-art alternatives on multiple real-world tasks, including digit/object recognition and sentiment analysis.} }
Endnote
%0 Conference Paper %T Domain Aggregation Networks for Multi-Source Domain Adaptation %A Junfeng Wen %A Russell Greiner %A Dale Schuurmans %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-wen20b %I PMLR %P 10214--10224 %U http://proceedings.mlr.press/v119/wen20b.html %V 119 %X In many real-world applications, we want to exploit multiple source datasets to build a model for a different but related target dataset. Despite the recent empirical success, most existing research has used ad-hoc methods to combine multiple sources, leading to a gap between theory and practice. In this paper, we develop a finite-sample generalization bound based on domain discrepancy and accordingly propose a theoretically justified optimization procedure. Our algorithm, Domain AggRegation Network (DARN), can automatically and dynamically balance between including more data to increase effective sample size and excluding irrelevant data to avoid negative effects during training. We find that DARN can significantly outperform the state-of-the-art alternatives on multiple real-world tasks, including digit/object recognition and sentiment analysis.
APA
Wen, J., Greiner, R. & Schuurmans, D.. (2020). Domain Aggregation Networks for Multi-Source Domain Adaptation. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:10214-10224 Available from http://proceedings.mlr.press/v119/wen20b.html.

Related Material