Learning from Multiple Sources via Multiple Domain Relationship

Zhen LIU; Junan YANG; Hui LIU; Jian LIU

doi:10.1587/transinf.2016EDL8008

Learning from Multiple Sources via Multiple Domain Relationship

Zhen LIU, Junan YANG, Hui LIU, Jian LIU

Full Text Views

0

Share
Cite this

Summary :

Transfer learning extracts useful information from the related source domain and leverages it to promote the target learning. The effectiveness of the transfer was affected by the relationship among domains. In this paper, a novel multi-source transfer learning based on multi-similarity was proposed. The method could increase the chance of finding the sources closely related to the target to reduce the “negative transfer” and also import more knowledge from multiple sources for the target learning. The method explored the relationship between the sources and the target by multi-similarity metric. Then, the knowledge of the sources was transferred to the target based on the smoothness assumption, which enforced that the target classifier shares similar decision values with the relevant source classifiers on the unlabeled target samples. Experimental results demonstrate that the proposed method can more effectively enhance the learning performance.

Publication: IEICE TRANSACTIONS on Information Vol.E99-D No.7 pp.1941-1944

Publication Date: 2016/07/01

Publicized: 2016/04/11

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDL8008

Type of Manuscript: LETTER

Category: Pattern Recognition

Authors

Zhen LIU
  Electronic Engineering Institution
Junan YANG
  Electronic Engineering Institution
Hui LIU
  Electronic Engineering Institution
Jian LIU
  Nanjing University of Posts and Telecommunications

Keyword

transfer learning, multiple source transfer, domain similarity, manifold assumption

Cite this

Copy

Zhen LIU, Junan YANG, Hui LIU, Jian LIU, "Learning from Multiple Sources via Multiple Domain Relationship" in IEICE TRANSACTIONS on Information, vol. E99-D, no. 7, pp. 1941-1944, July 2016, doi: 10.1587/transinf.2016EDL8008.
Abstract: Transfer learning extracts useful information from the related source domain and leverages it to promote the target learning. The effectiveness of the transfer was affected by the relationship among domains. In this paper, a novel multi-source transfer learning based on multi-similarity was proposed. The method could increase the chance of finding the sources closely related to the target to reduce the “negative transfer” and also import more knowledge from multiple sources for the target learning. The method explored the relationship between the sources and the target by multi-similarity metric. Then, the knowledge of the sources was transferred to the target based on the smoothness assumption, which enforced that the target classifier shares similar decision values with the relevant source classifiers on the unlabeled target samples. Experimental results demonstrate that the proposed method can more effectively enhance the learning performance.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2016EDL8008/_p

Copy

@ARTICLE{e99-d_7_1941,
author={Zhen LIU, Junan YANG, Hui LIU, Jian LIU, },
journal={IEICE TRANSACTIONS on Information},
title={Learning from Multiple Sources via Multiple Domain Relationship},
year={2016},
volume={E99-D},
number={7},
pages={1941-1944},
abstract={Transfer learning extracts useful information from the related source domain and leverages it to promote the target learning. The effectiveness of the transfer was affected by the relationship among domains. In this paper, a novel multi-source transfer learning based on multi-similarity was proposed. The method could increase the chance of finding the sources closely related to the target to reduce the “negative transfer” and also import more knowledge from multiple sources for the target learning. The method explored the relationship between the sources and the target by multi-similarity metric. Then, the knowledge of the sources was transferred to the target based on the smoothness assumption, which enforced that the target classifier shares similar decision values with the relevant source classifiers on the unlabeled target samples. Experimental results demonstrate that the proposed method can more effectively enhance the learning performance.},
keywords={},
doi={10.1587/transinf.2016EDL8008},
ISSN={1745-1361},
month={July},}

Copy

TY - JOUR
TI - Learning from Multiple Sources via Multiple Domain Relationship
T2 - IEICE TRANSACTIONS on Information
SP - 1941
EP - 1944
AU - Zhen LIU
AU - Junan YANG
AU - Hui LIU
AU - Jian LIU
PY - 2016
DO - 10.1587/transinf.2016EDL8008
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2016
AB - Transfer learning extracts useful information from the related source domain and leverages it to promote the target learning. The effectiveness of the transfer was affected by the relationship among domains. In this paper, a novel multi-source transfer learning based on multi-similarity was proposed. The method could increase the chance of finding the sources closely related to the target to reduce the “negative transfer” and also import more knowledge from multiple sources for the target learning. The method explored the relationship between the sources and the target by multi-similarity metric. Then, the knowledge of the sources was transferred to the target based on the smoothness assumption, which enforced that the target classifier shares similar decision values with the relevant source classifiers on the unlabeled target samples. Experimental results demonstrate that the proposed method can more effectively enhance the learning performance.
ER -