Adversarial domain adaptive subspace clustering

Mahdi Abavisani, Vishal Patel

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

We propose a novel method for clustering a collection of data that comes from several domains. Since members of the same class might look very different across different domains and because in a clustering problem we have no side information such as labels, the main challenge in a domain adaptive clustering problem is to group the data into different clusters regardless of their domain. We approach this problem by finding mappings that can transfer the data points between the domains. We use adversarial networks to approximate these mapping functions, and form a paired representation at each domain by mapping the data onto their counter domains. Finally, we employ a multimodal subspace clustering type algorithm to cluster the paired representations with respect to their subspaces. Various experiments on datasets with domain shifts show that our method performs significantly better than many competitive domain adaptive subspace clustering methods.

Original languageEnglish (US)
Title of host publication2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1-8
Number of pages8
ISBN (Electronic)9781538622483
DOIs
StatePublished - Mar 9 2018
Event4th IEEE International Conference on Identity, Security, and Behavior Analysis, ISBA 2018 - Singapore, Singapore
Duration: Jan 11 2018Jan 12 2018

Publication series

Name2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018
Volume2018-January

Other

Other4th IEEE International Conference on Identity, Security, and Behavior Analysis, ISBA 2018
CountrySingapore
CitySingapore
Period1/11/181/12/18

Fingerprint

Cluster Analysis
Labels
experiment
Experiments
Group

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Computer Vision and Pattern Recognition
  • Safety, Risk, Reliability and Quality
  • Behavioral Neuroscience
  • Social Sciences (miscellaneous)

Cite this

Abavisani, M., & Patel, V. (2018). Adversarial domain adaptive subspace clustering. In 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018 (pp. 1-8). (2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018; Vol. 2018-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISBA.2018.8311478
Abavisani, Mahdi ; Patel, Vishal. / Adversarial domain adaptive subspace clustering. 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 1-8 (2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018).
@inproceedings{3f98be70621249d187628487eed7ad7d,
title = "Adversarial domain adaptive subspace clustering",
abstract = "We propose a novel method for clustering a collection of data that comes from several domains. Since members of the same class might look very different across different domains and because in a clustering problem we have no side information such as labels, the main challenge in a domain adaptive clustering problem is to group the data into different clusters regardless of their domain. We approach this problem by finding mappings that can transfer the data points between the domains. We use adversarial networks to approximate these mapping functions, and form a paired representation at each domain by mapping the data onto their counter domains. Finally, we employ a multimodal subspace clustering type algorithm to cluster the paired representations with respect to their subspaces. Various experiments on datasets with domain shifts show that our method performs significantly better than many competitive domain adaptive subspace clustering methods.",
author = "Mahdi Abavisani and Vishal Patel",
year = "2018",
month = "3",
day = "9",
doi = "10.1109/ISBA.2018.8311478",
language = "English (US)",
series = "2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "1--8",
booktitle = "2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018",
address = "United States",

}

Abavisani, M & Patel, V 2018, Adversarial domain adaptive subspace clustering. in 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018. 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018, vol. 2018-January, Institute of Electrical and Electronics Engineers Inc., pp. 1-8, 4th IEEE International Conference on Identity, Security, and Behavior Analysis, ISBA 2018, Singapore, Singapore, 1/11/18. https://doi.org/10.1109/ISBA.2018.8311478

Adversarial domain adaptive subspace clustering. / Abavisani, Mahdi; Patel, Vishal.

2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 1-8 (2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018; Vol. 2018-January).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Adversarial domain adaptive subspace clustering

AU - Abavisani, Mahdi

AU - Patel, Vishal

PY - 2018/3/9

Y1 - 2018/3/9

N2 - We propose a novel method for clustering a collection of data that comes from several domains. Since members of the same class might look very different across different domains and because in a clustering problem we have no side information such as labels, the main challenge in a domain adaptive clustering problem is to group the data into different clusters regardless of their domain. We approach this problem by finding mappings that can transfer the data points between the domains. We use adversarial networks to approximate these mapping functions, and form a paired representation at each domain by mapping the data onto their counter domains. Finally, we employ a multimodal subspace clustering type algorithm to cluster the paired representations with respect to their subspaces. Various experiments on datasets with domain shifts show that our method performs significantly better than many competitive domain adaptive subspace clustering methods.

AB - We propose a novel method for clustering a collection of data that comes from several domains. Since members of the same class might look very different across different domains and because in a clustering problem we have no side information such as labels, the main challenge in a domain adaptive clustering problem is to group the data into different clusters regardless of their domain. We approach this problem by finding mappings that can transfer the data points between the domains. We use adversarial networks to approximate these mapping functions, and form a paired representation at each domain by mapping the data onto their counter domains. Finally, we employ a multimodal subspace clustering type algorithm to cluster the paired representations with respect to their subspaces. Various experiments on datasets with domain shifts show that our method performs significantly better than many competitive domain adaptive subspace clustering methods.

UR - http://www.scopus.com/inward/record.url?scp=85049810346&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049810346&partnerID=8YFLogxK

U2 - 10.1109/ISBA.2018.8311478

DO - 10.1109/ISBA.2018.8311478

M3 - Conference contribution

AN - SCOPUS:85049810346

T3 - 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018

SP - 1

EP - 8

BT - 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Abavisani M, Patel V. Adversarial domain adaptive subspace clustering. In 2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 1-8. (2018 IEEE 4th International Conference on Identity, Security, and Behavior Analysis, ISBA 2018). https://doi.org/10.1109/ISBA.2018.8311478