Cross-domain neural-kernel networks

Siamak Mehrkanoon

doi:10.1016/j.patrec.2019.06.007

Cross-domain neural-kernel networks

Siamak Mehrkanoon

extern

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

This paper introduces a novel cross-domain neural-kernel networks architecture for semi-supervised domain adaption problem. The proposed model consists of two stream neural-kernel networks corresponding to the source and target domains which are enriched with a coupling term. Each stream neural-kernel networks follows a combination of neural network layer and an explicit feature map constructed by means of random Fourier features. The introduced coupling term aims at enforcing correlations among the output of the intermediate layers of the two stream networks as well as encouraging the two networks to learn shared representation of the data from both source and target domains. Experimental results are given to illustrate the effectiveness of the proposed approaches on synthetic and real-life datasets.

Original language	English
Pages (from-to)	474-480
Number of pages	7
Journal	Pattern Recognition Letters
Volume	125
DOIs	https://doi.org/10.1016/j.patrec.2019.06.007
Publication status	Published - 1 Jul 2019

Bibliographical note

Funding Information:
This work was partially supported by the Postdoctoral Fellowship of the Research Foundation -Flanders (FWO: 12Z1318N ). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands.

Funding Information:
This work was partially supported by the Postdoctoral Fellowship of the Research Foundation-Flanders (FWO: 12Z1318N). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands.

Publisher Copyright:
© 2019 Elsevier B.V.

Funding

This work was partially supported by the Postdoctoral Fellowship of the Research Foundation -Flanders (FWO: 12Z1318N ). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands. This work was partially supported by the Postdoctoral Fellowship of the Research Foundation-Flanders (FWO: 12Z1318N). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands.

Keywords

Coupling regularization
Domain adaptation
Kernel methods
Neural networks

Access to Document

10.1016/j.patrec.2019.06.007

Cite this

@article{9337e68315624f60b417f3d62eafd83a,

title = "Cross-domain neural-kernel networks",

abstract = "This paper introduces a novel cross-domain neural-kernel networks architecture for semi-supervised domain adaption problem. The proposed model consists of two stream neural-kernel networks corresponding to the source and target domains which are enriched with a coupling term. Each stream neural-kernel networks follows a combination of neural network layer and an explicit feature map constructed by means of random Fourier features. The introduced coupling term aims at enforcing correlations among the output of the intermediate layers of the two stream networks as well as encouraging the two networks to learn shared representation of the data from both source and target domains. Experimental results are given to illustrate the effectiveness of the proposed approaches on synthetic and real-life datasets.",

keywords = "Coupling regularization, Domain adaptation, Kernel methods, Neural networks",

author = "Siamak Mehrkanoon",

note = "Funding Information: This work was partially supported by the Postdoctoral Fellowship of the Research Foundation -Flanders (FWO: 12Z1318N ). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands. Funding Information: This work was partially supported by the Postdoctoral Fellowship of the Research Foundation-Flanders (FWO: 12Z1318N). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands. Publisher Copyright: {\textcopyright} 2019 Elsevier B.V.",

year = "2019",

month = jul,

day = "1",

doi = "10.1016/j.patrec.2019.06.007",

language = "English",

volume = "125",

pages = "474--480",

journal = "Pattern Recognition Letters",

issn = "0167-8655",

publisher = "Elsevier BV",

}

TY - JOUR

T1 - Cross-domain neural-kernel networks

AU - Mehrkanoon, Siamak

N1 - Funding Information: This work was partially supported by the Postdoctoral Fellowship of the Research Foundation -Flanders (FWO: 12Z1318N ). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands. Funding Information: This work was partially supported by the Postdoctoral Fellowship of the Research Foundation-Flanders (FWO: 12Z1318N). Siamak Mehrkanoon is an assistant professor at the Department of Data Science and Knowledge Engineering, Maastricht University, The Netherlands. Publisher Copyright: © 2019 Elsevier B.V.

PY - 2019/7/1

Y1 - 2019/7/1

N2 - This paper introduces a novel cross-domain neural-kernel networks architecture for semi-supervised domain adaption problem. The proposed model consists of two stream neural-kernel networks corresponding to the source and target domains which are enriched with a coupling term. Each stream neural-kernel networks follows a combination of neural network layer and an explicit feature map constructed by means of random Fourier features. The introduced coupling term aims at enforcing correlations among the output of the intermediate layers of the two stream networks as well as encouraging the two networks to learn shared representation of the data from both source and target domains. Experimental results are given to illustrate the effectiveness of the proposed approaches on synthetic and real-life datasets.

AB - This paper introduces a novel cross-domain neural-kernel networks architecture for semi-supervised domain adaption problem. The proposed model consists of two stream neural-kernel networks corresponding to the source and target domains which are enriched with a coupling term. Each stream neural-kernel networks follows a combination of neural network layer and an explicit feature map constructed by means of random Fourier features. The introduced coupling term aims at enforcing correlations among the output of the intermediate layers of the two stream networks as well as encouraging the two networks to learn shared representation of the data from both source and target domains. Experimental results are given to illustrate the effectiveness of the proposed approaches on synthetic and real-life datasets.

KW - Coupling regularization

KW - Domain adaptation

KW - Kernel methods

KW - Neural networks

UR - http://www.scopus.com/inward/record.url?scp=85067235783&partnerID=8YFLogxK

U2 - 10.1016/j.patrec.2019.06.007

DO - 10.1016/j.patrec.2019.06.007

M3 - Article

AN - SCOPUS:85067235783

SN - 0167-8655

VL - 125

SP - 474

EP - 480

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

ER -

Cross-domain neural-kernel networks

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this