TY - GEN

T1 - Domain adaptation with multiple sources

AU - Mansour, Yishay

AU - Mohri, Mehryar

AU - Rostamizadeh, Afshin

PY - 2009

Y1 - 2009

N2 - This paper presents a theoretical analysis of the problem of domain adaptation with multiple sources. For each source domain, the distribution over the input points as well as a hypothesis with error at most o are given. The problem consists of combining these hypotheses to derive a hypothesis with small error with respect to the target domain. We present several theoretical results relating to this problem. In particular, we prove that standard convex combinations of the source hypothesesmay in fact performvery poorly and that, instead, combinations weighted by the source distributions benefit from favorable theoretical guarantees. Our main result shows that, remarkably, for any fixed target function, there exists a distribution weighted combining rule that has a loss of at most ε with respect to any target mixture of the source distributions. We further generalize the setting from a single target function to multiple consistent target functions and show the existence of a combining rule with error at most 3ε. Finally, we report empirical results for a multiple source adaptation problem with a real-world dataset.

AB - This paper presents a theoretical analysis of the problem of domain adaptation with multiple sources. For each source domain, the distribution over the input points as well as a hypothesis with error at most o are given. The problem consists of combining these hypotheses to derive a hypothesis with small error with respect to the target domain. We present several theoretical results relating to this problem. In particular, we prove that standard convex combinations of the source hypothesesmay in fact performvery poorly and that, instead, combinations weighted by the source distributions benefit from favorable theoretical guarantees. Our main result shows that, remarkably, for any fixed target function, there exists a distribution weighted combining rule that has a loss of at most ε with respect to any target mixture of the source distributions. We further generalize the setting from a single target function to multiple consistent target functions and show the existence of a combining rule with error at most 3ε. Finally, we report empirical results for a multiple source adaptation problem with a real-world dataset.

UR - http://www.scopus.com/inward/record.url?scp=70049090062&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70049090062&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:70049090062

SN - 9781605609492

T3 - Advances in Neural Information Processing Systems 21 - Proceedings of the 2008 Conference

SP - 1041

EP - 1048

BT - Advances in Neural Information Processing Systems 21 - Proceedings of the 2008 Conference

T2 - 22nd Annual Conference on Neural Information Processing Systems, NIPS 2008

Y2 - 8 December 2008 through 11 December 2008

ER -