SUBSCALE: Fast and scalable subspace clustering for high dimensional data

Amardeep Kaur; Amitava Datta

doi:10.1109/ICDMW.2014.100

SUBSCALE: Fast and scalable subspace clustering for high dimensional data

Amardeep Kaur, Amitava Datta

Research output: Chapter in Book/Conference paper › Conference paper › peer-review

7 Citations (Scopus)

Abstract

© 2014 IEEE. The aim of subspace clustering is to find groups of similar data points in all possible subspaces of a dataset. Since the number of subspaces is exponential in dimensions, subspace clustering is usually computationally very expensive. The performance of existing algorithms deteriorates drastically with the increase in number of dimensions. Most of them use bottom-up search strategy and there are two main reasons for their inefficiency: (1) Multiple database scans. (2) Either implicit or explicit generation of trivial subspace clusters during the process. We present SUBSCALE, a novel algorithm to directly find the non-trivial subspace clusters with minimal cost and it requires only k database scans for a k-dimensional data set. Our algorithm scales very well with the dimensionality and is highly parallelizable. The experimental evaluation has shown promising results.

Original language	English
Title of host publication	Proceedings 2014 IEEE International Conference on Data Mining Workshop (ICDMW)
Place of Publication	New Jersey, USA
Publisher	IEEE, Institute of Electrical and Electronics Engineers
Pages	621-628
Volume	N/A
ISBN (Electronic)	9781479942756
ISBN (Print)	9781479942749
DOIs	https://doi.org/10.1109/ICDMW.2014.100
Publication status	Published - 2015
Event	2014 IEEE International Conference on Data Mining Workshop - Shenzhen, China Duration: 14 Dec 2014 → 14 Dec 2014

Workshop

Workshop	2014 IEEE International Conference on Data Mining Workshop
Abbreviated title	ICDMW
Country/Territory	China
City	Shenzhen
Period	14/12/14 → 14/12/14

Access to Document

10.1109/ICDMW.2014.100

Cite this

@inproceedings{b1e8c6b69f92427c983faf5ad18dc99d,

title = "SUBSCALE: Fast and scalable subspace clustering for high dimensional data",

abstract = "{\textcopyright} 2014 IEEE. The aim of subspace clustering is to find groups of similar data points in all possible subspaces of a dataset. Since the number of subspaces is exponential in dimensions, subspace clustering is usually computationally very expensive. The performance of existing algorithms deteriorates drastically with the increase in number of dimensions. Most of them use bottom-up search strategy and there are two main reasons for their inefficiency: (1) Multiple database scans. (2) Either implicit or explicit generation of trivial subspace clusters during the process. We present SUBSCALE, a novel algorithm to directly find the non-trivial subspace clusters with minimal cost and it requires only k database scans for a k-dimensional data set. Our algorithm scales very well with the dimensionality and is highly parallelizable. The experimental evaluation has shown promising results.",

author = "Amardeep Kaur and Amitava Datta",

year = "2015",

doi = "10.1109/ICDMW.2014.100",

language = "English",

isbn = "9781479942749",

volume = "N/A",

pages = "621--628",

booktitle = "Proceedings 2014 IEEE International Conference on Data Mining Workshop (ICDMW)",

publisher = "IEEE, Institute of Electrical and Electronics Engineers",

address = "United States",

note = "2014 IEEE International Conference on Data Mining Workshop, ICDMW ; Conference date: 14-12-2014 Through 14-12-2014",

}

Kaur, A & Datta, A 2015, SUBSCALE: Fast and scalable subspace clustering for high dimensional data. in Proceedings 2014 IEEE International Conference on Data Mining Workshop (ICDMW). vol. N/A, IEEE, Institute of Electrical and Electronics Engineers, New Jersey, USA, pp. 621-628, 2014 IEEE International Conference on Data Mining Workshop, Shenzhen, China, 14/12/14. https://doi.org/10.1109/ICDMW.2014.100

SUBSCALE: Fast and scalable subspace clustering for high dimensional data. / Kaur, Amardeep; Datta, Amitava.
Proceedings 2014 IEEE International Conference on Data Mining Workshop (ICDMW). Vol. N/A New Jersey, USA: IEEE, Institute of Electrical and Electronics Engineers, 2015. p. 621-628.

Research output: Chapter in Book/Conference paper › Conference paper › peer-review

TY - GEN

T1 - SUBSCALE: Fast and scalable subspace clustering for high dimensional data

AU - Kaur, Amardeep

AU - Datta, Amitava

PY - 2015

Y1 - 2015

N2 - © 2014 IEEE. The aim of subspace clustering is to find groups of similar data points in all possible subspaces of a dataset. Since the number of subspaces is exponential in dimensions, subspace clustering is usually computationally very expensive. The performance of existing algorithms deteriorates drastically with the increase in number of dimensions. Most of them use bottom-up search strategy and there are two main reasons for their inefficiency: (1) Multiple database scans. (2) Either implicit or explicit generation of trivial subspace clusters during the process. We present SUBSCALE, a novel algorithm to directly find the non-trivial subspace clusters with minimal cost and it requires only k database scans for a k-dimensional data set. Our algorithm scales very well with the dimensionality and is highly parallelizable. The experimental evaluation has shown promising results.

AB - © 2014 IEEE. The aim of subspace clustering is to find groups of similar data points in all possible subspaces of a dataset. Since the number of subspaces is exponential in dimensions, subspace clustering is usually computationally very expensive. The performance of existing algorithms deteriorates drastically with the increase in number of dimensions. Most of them use bottom-up search strategy and there are two main reasons for their inefficiency: (1) Multiple database scans. (2) Either implicit or explicit generation of trivial subspace clusters during the process. We present SUBSCALE, a novel algorithm to directly find the non-trivial subspace clusters with minimal cost and it requires only k database scans for a k-dimensional data set. Our algorithm scales very well with the dimensionality and is highly parallelizable. The experimental evaluation has shown promising results.

U2 - 10.1109/ICDMW.2014.100

DO - 10.1109/ICDMW.2014.100

M3 - Conference paper

SN - 9781479942749

VL - N/A

SP - 621

EP - 628

BT - Proceedings 2014 IEEE International Conference on Data Mining Workshop (ICDMW)

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - New Jersey, USA

T2 - 2014 IEEE International Conference on Data Mining Workshop

Y2 - 14 December 2014 through 14 December 2014

ER -

SUBSCALE: Fast and scalable subspace clustering for high dimensional data

Abstract

Workshop

Access to Document

Fingerprint

Cite this