Transfer learning by centroid pivoted mapping in noisy environment

Thach Nguyen Huy, Bin Tong, Hao Shao, Einoshin Suzuki

Research output: Contribution to journalArticle

Abstract

Transfer learning is a widely investigated learning paradigm that is initially proposed to reuse informative knowledge from related domains, as supervised information in the target domain is scarce while it is sufficiently available in the multiple source domains. One of the challenging issues in transfer learning is how to handle the distribution differences between the source domains and the target domain. Most studies in the research field implicitly assume that data distributions from the source domains and the target domain are similar in a well-designed feature space. However, it is often the case that label assignments for data in the source domains and the target domain are significantly different. Therefore, in reality even if the distribution difference between a source domain and a target domain is reduced, the knowledge from multiple source domains is not well transferred to the target domain unless the label information is carefully considered. In addition, noisy data often emerge in real world applications. Therefore, considering how to handle noisy data in the transfer learning setting is a challenging problem, as noisy data inevitably cause a side effect during the knowledge transfer. Due to the above reasons, in this paper, we are motivated to propose a robust framework against noise in the transfer learning setting. We also explicitly consider the difference in data distributions and label assignments among multiple source domains and the target domain. Experimental results on one synthetic data set, three UCI data sets and one real world text data set in different noise levels demonstrate the effectiveness of our method.

Original languageEnglish
Pages (from-to)39-60
Number of pages22
JournalJournal of Intelligent Information Systems
Volume41
Issue number1
DOIs
Publication statusPublished - Aug 1 2013

Fingerprint

Labels

All Science Journal Classification (ASJC) codes

  • Software
  • Information Systems
  • Hardware and Architecture
  • Computer Networks and Communications
  • Artificial Intelligence

Cite this

Transfer learning by centroid pivoted mapping in noisy environment. / Huy, Thach Nguyen; Tong, Bin; Shao, Hao; Suzuki, Einoshin.

In: Journal of Intelligent Information Systems, Vol. 41, No. 1, 01.08.2013, p. 39-60.

Research output: Contribution to journalArticle

Huy, Thach Nguyen ; Tong, Bin ; Shao, Hao ; Suzuki, Einoshin. / Transfer learning by centroid pivoted mapping in noisy environment. In: Journal of Intelligent Information Systems. 2013 ; Vol. 41, No. 1. pp. 39-60.
@article{6114b26b58554de7987d6de47738a40f,
title = "Transfer learning by centroid pivoted mapping in noisy environment",
abstract = "Transfer learning is a widely investigated learning paradigm that is initially proposed to reuse informative knowledge from related domains, as supervised information in the target domain is scarce while it is sufficiently available in the multiple source domains. One of the challenging issues in transfer learning is how to handle the distribution differences between the source domains and the target domain. Most studies in the research field implicitly assume that data distributions from the source domains and the target domain are similar in a well-designed feature space. However, it is often the case that label assignments for data in the source domains and the target domain are significantly different. Therefore, in reality even if the distribution difference between a source domain and a target domain is reduced, the knowledge from multiple source domains is not well transferred to the target domain unless the label information is carefully considered. In addition, noisy data often emerge in real world applications. Therefore, considering how to handle noisy data in the transfer learning setting is a challenging problem, as noisy data inevitably cause a side effect during the knowledge transfer. Due to the above reasons, in this paper, we are motivated to propose a robust framework against noise in the transfer learning setting. We also explicitly consider the difference in data distributions and label assignments among multiple source domains and the target domain. Experimental results on one synthetic data set, three UCI data sets and one real world text data set in different noise levels demonstrate the effectiveness of our method.",
author = "Huy, {Thach Nguyen} and Bin Tong and Hao Shao and Einoshin Suzuki",
year = "2013",
month = "8",
day = "1",
doi = "10.1007/s10844-012-0226-3",
language = "English",
volume = "41",
pages = "39--60",
journal = "Journal of Intelligent Information Systems",
issn = "0925-9902",
publisher = "Springer Netherlands",
number = "1",

}

TY - JOUR

T1 - Transfer learning by centroid pivoted mapping in noisy environment

AU - Huy, Thach Nguyen

AU - Tong, Bin

AU - Shao, Hao

AU - Suzuki, Einoshin

PY - 2013/8/1

Y1 - 2013/8/1

N2 - Transfer learning is a widely investigated learning paradigm that is initially proposed to reuse informative knowledge from related domains, as supervised information in the target domain is scarce while it is sufficiently available in the multiple source domains. One of the challenging issues in transfer learning is how to handle the distribution differences between the source domains and the target domain. Most studies in the research field implicitly assume that data distributions from the source domains and the target domain are similar in a well-designed feature space. However, it is often the case that label assignments for data in the source domains and the target domain are significantly different. Therefore, in reality even if the distribution difference between a source domain and a target domain is reduced, the knowledge from multiple source domains is not well transferred to the target domain unless the label information is carefully considered. In addition, noisy data often emerge in real world applications. Therefore, considering how to handle noisy data in the transfer learning setting is a challenging problem, as noisy data inevitably cause a side effect during the knowledge transfer. Due to the above reasons, in this paper, we are motivated to propose a robust framework against noise in the transfer learning setting. We also explicitly consider the difference in data distributions and label assignments among multiple source domains and the target domain. Experimental results on one synthetic data set, three UCI data sets and one real world text data set in different noise levels demonstrate the effectiveness of our method.

AB - Transfer learning is a widely investigated learning paradigm that is initially proposed to reuse informative knowledge from related domains, as supervised information in the target domain is scarce while it is sufficiently available in the multiple source domains. One of the challenging issues in transfer learning is how to handle the distribution differences between the source domains and the target domain. Most studies in the research field implicitly assume that data distributions from the source domains and the target domain are similar in a well-designed feature space. However, it is often the case that label assignments for data in the source domains and the target domain are significantly different. Therefore, in reality even if the distribution difference between a source domain and a target domain is reduced, the knowledge from multiple source domains is not well transferred to the target domain unless the label information is carefully considered. In addition, noisy data often emerge in real world applications. Therefore, considering how to handle noisy data in the transfer learning setting is a challenging problem, as noisy data inevitably cause a side effect during the knowledge transfer. Due to the above reasons, in this paper, we are motivated to propose a robust framework against noise in the transfer learning setting. We also explicitly consider the difference in data distributions and label assignments among multiple source domains and the target domain. Experimental results on one synthetic data set, three UCI data sets and one real world text data set in different noise levels demonstrate the effectiveness of our method.

UR - http://www.scopus.com/inward/record.url?scp=84882249714&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84882249714&partnerID=8YFLogxK

U2 - 10.1007/s10844-012-0226-3

DO - 10.1007/s10844-012-0226-3

M3 - Article

AN - SCOPUS:84882249714

VL - 41

SP - 39

EP - 60

JO - Journal of Intelligent Information Systems

JF - Journal of Intelligent Information Systems

SN - 0925-9902

IS - 1

ER -