Undercomplete blind subspace deconvolution via linear prediction

Zoltán Szabó, Barnabás Póczos, A. Lőrincz

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

5 Citations (Scopus)

Abstract

We present a novel solution technique for the blind subspace deconvolution (BSSD) problem, where temporal convolution of multidimensional hidden independent components is observed and the task is to uncover the hidden components using the observation only. We carry out this task for the undercomplete case (uBSSD): we reduce the original uBSSD task via linear prediction to independent subspace analysis (ISA), which we can solve. As it has been shown recently, applying temporal concatenation can also reduce uBSSD to ISA, but the associated ISA problem can easily become 'high dimensional' [1]. The new reduction method circumvents this dimensionality problem. We perform detailed studies on the efficiency of the proposed technique by means of numerical simulations. We have found several advantages: our method can achieve high quality estimations for smaller number of samples and it can cope with deeper temporal convolutions.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages740-747
Number of pages8
Volume4701 LNAI
Publication statusPublished - 2007
Event18th European Conference on Machine Learning, ECML 2007 - Warsaw, Poland
Duration: Sep 17 2007Sep 21 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4701 LNAI
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other18th European Conference on Machine Learning, ECML 2007
CountryPoland
CityWarsaw
Period9/17/079/21/07

Fingerprint

Linear Prediction
Deconvolution
Convolution
Subspace
Concatenation
Computer simulation
Reduction Method
Observation
Dimensionality
High-dimensional
Numerical Simulation

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Cite this

Szabó, Z., Póczos, B., & Lőrincz, A. (2007). Undercomplete blind subspace deconvolution via linear prediction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4701 LNAI, pp. 740-747). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4701 LNAI).

Undercomplete blind subspace deconvolution via linear prediction. / Szabó, Zoltán; Póczos, Barnabás; Lőrincz, A.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4701 LNAI 2007. p. 740-747 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4701 LNAI).

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

Szabó, Z, Póczos, B & Lőrincz, A 2007, Undercomplete blind subspace deconvolution via linear prediction. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 4701 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4701 LNAI, pp. 740-747, 18th European Conference on Machine Learning, ECML 2007, Warsaw, Poland, 9/17/07.
Szabó Z, Póczos B, Lőrincz A. Undercomplete blind subspace deconvolution via linear prediction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4701 LNAI. 2007. p. 740-747. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
Szabó, Zoltán ; Póczos, Barnabás ; Lőrincz, A. / Undercomplete blind subspace deconvolution via linear prediction. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4701 LNAI 2007. pp. 740-747 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{e3b09837ff434a6da1466e9934cf9261,
title = "Undercomplete blind subspace deconvolution via linear prediction",
abstract = "We present a novel solution technique for the blind subspace deconvolution (BSSD) problem, where temporal convolution of multidimensional hidden independent components is observed and the task is to uncover the hidden components using the observation only. We carry out this task for the undercomplete case (uBSSD): we reduce the original uBSSD task via linear prediction to independent subspace analysis (ISA), which we can solve. As it has been shown recently, applying temporal concatenation can also reduce uBSSD to ISA, but the associated ISA problem can easily become 'high dimensional' [1]. The new reduction method circumvents this dimensionality problem. We perform detailed studies on the efficiency of the proposed technique by means of numerical simulations. We have found several advantages: our method can achieve high quality estimations for smaller number of samples and it can cope with deeper temporal convolutions.",
author = "Zolt{\'a}n Szab{\'o} and Barnab{\'a}s P{\'o}czos and A. Lőrincz",
year = "2007",
language = "English",
isbn = "9783540749578",
volume = "4701 LNAI",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "740--747",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Undercomplete blind subspace deconvolution via linear prediction

AU - Szabó, Zoltán

AU - Póczos, Barnabás

AU - Lőrincz, A.

PY - 2007

Y1 - 2007

N2 - We present a novel solution technique for the blind subspace deconvolution (BSSD) problem, where temporal convolution of multidimensional hidden independent components is observed and the task is to uncover the hidden components using the observation only. We carry out this task for the undercomplete case (uBSSD): we reduce the original uBSSD task via linear prediction to independent subspace analysis (ISA), which we can solve. As it has been shown recently, applying temporal concatenation can also reduce uBSSD to ISA, but the associated ISA problem can easily become 'high dimensional' [1]. The new reduction method circumvents this dimensionality problem. We perform detailed studies on the efficiency of the proposed technique by means of numerical simulations. We have found several advantages: our method can achieve high quality estimations for smaller number of samples and it can cope with deeper temporal convolutions.

AB - We present a novel solution technique for the blind subspace deconvolution (BSSD) problem, where temporal convolution of multidimensional hidden independent components is observed and the task is to uncover the hidden components using the observation only. We carry out this task for the undercomplete case (uBSSD): we reduce the original uBSSD task via linear prediction to independent subspace analysis (ISA), which we can solve. As it has been shown recently, applying temporal concatenation can also reduce uBSSD to ISA, but the associated ISA problem can easily become 'high dimensional' [1]. The new reduction method circumvents this dimensionality problem. We perform detailed studies on the efficiency of the proposed technique by means of numerical simulations. We have found several advantages: our method can achieve high quality estimations for smaller number of samples and it can cope with deeper temporal convolutions.

UR - http://www.scopus.com/inward/record.url?scp=38049132542&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38049132542&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540749578

VL - 4701 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 740

EP - 747

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -