Nonlinear Blind Source Separation Unifying Vanishing Component Analysis and Temporal Structure

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Nonlinear blind source separation (BSS) is one of the unsolved problems in unsupervised learning, because the solutions are highly non-unique when there is no prior information for the mixing functions. In this paper, we present a novel approach to tackle the ill-posedness of the nonlinear BSS problem with a few assumptions. The derivation of our algorithm is inspired by the idea of an efficient layerby- layer representation to approximate the nonlinearity. Once such a representation is built, a final output layer is constructed by solving a convex optimization problem. Relying on the multi-layer architecture, the algorithm transforms a time-invariant nonlinear BSS to the local linear problem with a tolerable computational cost. Then the projected data can break the nonlinear problem down into the version of a generalized joint diagonalization problem in the feature space. Importantly, the parameters and forms of polynomials depend solely on the input data, which guarantee the robustness of the structure. We thus address the general problem without being restricted to any specific mixture or parametric model. Experiments show that the proposed algorithm has a higher estimation accuracy on audio datasets from the real world for separating various nonlinear mixtures.

Original languageEnglish
JournalIEEE Access
DOIs
Publication statusAccepted/In press - 2018 Jul 28

Fingerprint

Blind source separation
Unsupervised learning
Convex optimization
Polynomials
Costs
Experiments

Keywords

  • Nonlinear BSS
  • statistical independent
  • temporal structure
  • vanishing component analysis

ASJC Scopus subject areas

  • Computer Science(all)
  • Materials Science(all)
  • Engineering(all)

Cite this

@article{d29030f985a140668fe329d85a6b9754,
title = "Nonlinear Blind Source Separation Unifying Vanishing Component Analysis and Temporal Structure",
abstract = "Nonlinear blind source separation (BSS) is one of the unsolved problems in unsupervised learning, because the solutions are highly non-unique when there is no prior information for the mixing functions. In this paper, we present a novel approach to tackle the ill-posedness of the nonlinear BSS problem with a few assumptions. The derivation of our algorithm is inspired by the idea of an efficient layerby- layer representation to approximate the nonlinearity. Once such a representation is built, a final output layer is constructed by solving a convex optimization problem. Relying on the multi-layer architecture, the algorithm transforms a time-invariant nonlinear BSS to the local linear problem with a tolerable computational cost. Then the projected data can break the nonlinear problem down into the version of a generalized joint diagonalization problem in the feature space. Importantly, the parameters and forms of polynomials depend solely on the input data, which guarantee the robustness of the structure. We thus address the general problem without being restricted to any specific mixture or parametric model. Experiments show that the proposed algorithm has a higher estimation accuracy on audio datasets from the real world for separating various nonlinear mixtures.",
keywords = "Nonlinear BSS, statistical independent, temporal structure, vanishing component analysis",
author = "Lu Wang and Tomoaki Ohtsuki",
year = "2018",
month = "7",
day = "28",
doi = "10.1109/ACCESS.2018.2860683",
language = "English",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Nonlinear Blind Source Separation Unifying Vanishing Component Analysis and Temporal Structure

AU - Wang, Lu

AU - Ohtsuki, Tomoaki

PY - 2018/7/28

Y1 - 2018/7/28

N2 - Nonlinear blind source separation (BSS) is one of the unsolved problems in unsupervised learning, because the solutions are highly non-unique when there is no prior information for the mixing functions. In this paper, we present a novel approach to tackle the ill-posedness of the nonlinear BSS problem with a few assumptions. The derivation of our algorithm is inspired by the idea of an efficient layerby- layer representation to approximate the nonlinearity. Once such a representation is built, a final output layer is constructed by solving a convex optimization problem. Relying on the multi-layer architecture, the algorithm transforms a time-invariant nonlinear BSS to the local linear problem with a tolerable computational cost. Then the projected data can break the nonlinear problem down into the version of a generalized joint diagonalization problem in the feature space. Importantly, the parameters and forms of polynomials depend solely on the input data, which guarantee the robustness of the structure. We thus address the general problem without being restricted to any specific mixture or parametric model. Experiments show that the proposed algorithm has a higher estimation accuracy on audio datasets from the real world for separating various nonlinear mixtures.

AB - Nonlinear blind source separation (BSS) is one of the unsolved problems in unsupervised learning, because the solutions are highly non-unique when there is no prior information for the mixing functions. In this paper, we present a novel approach to tackle the ill-posedness of the nonlinear BSS problem with a few assumptions. The derivation of our algorithm is inspired by the idea of an efficient layerby- layer representation to approximate the nonlinearity. Once such a representation is built, a final output layer is constructed by solving a convex optimization problem. Relying on the multi-layer architecture, the algorithm transforms a time-invariant nonlinear BSS to the local linear problem with a tolerable computational cost. Then the projected data can break the nonlinear problem down into the version of a generalized joint diagonalization problem in the feature space. Importantly, the parameters and forms of polynomials depend solely on the input data, which guarantee the robustness of the structure. We thus address the general problem without being restricted to any specific mixture or parametric model. Experiments show that the proposed algorithm has a higher estimation accuracy on audio datasets from the real world for separating various nonlinear mixtures.

KW - Nonlinear BSS

KW - statistical independent

KW - temporal structure

KW - vanishing component analysis

UR - http://www.scopus.com/inward/record.url?scp=85050757974&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85050757974&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2018.2860683

DO - 10.1109/ACCESS.2018.2860683

M3 - Article

AN - SCOPUS:85050757974

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

ER -