Domain Generalization Based on Transfer Component Analysis

Grubinger, Thomas; Birlutiu, Adriana; Schöner, Holger; Natschläger, Thomas; Heskes, Tom

by Thomas Grubinger, Adriana Birlutiu, Holger Schöner, Thomas Natschläger, Tom Heskes

Abstract:

This paper investigates domain generalization: How to use knowledge acquired from related domains and apply it to new domains? Transfer Component Analysis (TCA) learns a shared subspace by minimizing the dissimilarities across domains, while maximally preserving the data variance. We propose Multi-TCA, an extension of TCA to multiple domains as well as Multi-SSTCA, which is an extension of TCA for semi-supervised learning. In addition to the original application of TCA for domain adaptation problems, we show that Multi-TCA can also be applied for domain generalization. Multi-TCA and Multi-SSTCA are evaluated on two publicly available datasets with the tasks of landmine detection and Parkinson telemonitoring. Experimental results demonstrate that Multi-TCA can improve predictive performance on previously unseen domains.

Reference:

Domain Generalization Based on Transfer Component Analysis (Thomas Grubinger, Adriana Birlutiu, Holger Schöner, Thomas Natschläger, Tom Heskes), In Advances in Computational Intelligence - Proceedings of IWANN 2015, Part I (I. Rojas, G. Joya, A. Catala, eds.), Springer, volume 9094, 2015.

Bibtex Entry:

@inproceedings{grubinger_domain_2015,
	title = {Domain {Generalization} {Based} on {Transfer} {Component} {Analysis}},
	volume = {9094},
	issn = {978-3-319-19257-4},
	abstract = {This paper investigates domain generalization: How to use knowledge acquired from related domains and apply it to new domains? Transfer Component Analysis (TCA) learns a shared subspace by minimizing the dissimilarities across domains, while maximally preserving the data variance. We propose Multi-TCA, an extension of TCA to multiple domains as well as Multi-SSTCA, which is an extension of TCA for semi-supervised learning. In addition to the original application of TCA for domain adaptation problems, we show that Multi-TCA can also be applied for domain generalization. Multi-TCA and Multi-SSTCA are evaluated on two publicly available datasets with the tasks of landmine detection and Parkinson telemonitoring. Experimental results demonstrate that Multi-TCA can improve predictive performance on previously unseen domains.},
	booktitle = {Advances in Computational Intelligence - Proceedings of IWANN 2015, Part I},
	author = {Grubinger, Thomas and Birlutiu, Adriana and Schöner, Holger and Natschläger, Thomas and Heskes, Tom},
	editor = {Rojas, I. and Joya, G. and Catala, A.},
	month = jun,
	year = {2015},
	publisher = {Springer},
	pages = {325--334},
	file = {download/publications/Grubinger2015_DomainGeneralizationBasedOnTransferComponentAnalysis.pdf}
}