Prototypes and matrix relevance learning in complex fourier space

M. Straat; M. Kaden; M. Gay; T. Villmann; Alexander Lampe; U. Seiffert; M. Biehl; F. Melchert

doi:10.1109/WSOM.2017.8020019

Prototypes and matrix relevance learning in complex fourier space

M. Straat, M. Kaden, M. Gay, T. Villmann, Alexander Lampe, U. Seiffert, M. Biehl, F. Melchert

Intelligente Systemen

Onderzoeksoutput › Academic › peer review

18 Citaten (Scopus)

32 Downloads (Pure)

Samenvatting

In this contribution, we consider the classification of time-series and similar functional data which can be represented in complex Fourier coefficient space. We apply versions of Learning Vector Quantization (LVQ) which are suitable for complex-valued data, based on the so-called Wirtinger calculus. It makes possible the formulation of gradient based update rules in the framework of cost-function based Generalized Matrix Relevance LVQ (GMLVQ). Alternatively, we consider the concatenation of real and imaginary parts of Fourier coefficients in a real-valued feature vector and the classification of time domain representations by means of conventional GMLVQ.

Originele taal-2	English
Titel	12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)
Uitgeverij	IEEEXplore
Pagina's	1-6
Aantal pagina's	6
ISBN van elektronische versie	978-1-5090-6638-4
DOI's	https://doi.org/10.1109/WSOM.2017.8020019
Status	Published - 31-aug.-2017
Evenement	12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM) - Nancy, France Duur: 28-jun.-2017 → 30-jun.-2017

Conference

Conference	12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)
Land/Regio	France
Stad	Nancy
Periode	28/06/2017 → 30/06/2017

Toegang tot document

10.1109/WSOM.2017.8020019

Prototypes and Matrix RelevanceFinal publisher's version, 464 KBLicentie: Taverne

Handle.net

Permanente koppeling

1 Attending an event
1 Academic presentation

Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization
Friedrich Melchert (Speaker), Michael LeKander (Speaker) & Michiel Straat (Speaker)
28-jun.-2017 → 30-jun.-2017
Activiteit: Academic presentation › Academic
Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization
Michael Biehl (Attendee)
28-jun.-2017 → 30-jun.-2017
Activiteit: Attending an event › Academic

Citeer dit

@inproceedings{20e9e2805bd34c0f831b034b90293424,

title = "Prototypes and matrix relevance learning in complex fourier space",

abstract = "In this contribution, we consider the classification of time-series and similar functional data which can be represented in complex Fourier coefficient space. We apply versions of Learning Vector Quantization (LVQ) which are suitable for complex-valued data, based on the so-called Wirtinger calculus. It makes possible the formulation of gradient based update rules in the framework of cost-function based Generalized Matrix Relevance LVQ (GMLVQ). Alternatively, we consider the concatenation of real and imaginary parts of Fourier coefficients in a real-valued feature vector and the classification of time domain representations by means of conventional GMLVQ.",

keywords = "Calculus, Discrete Fourier transforms, Frequency-domain analysis, Prototypes, Time series analysis, Time-domain analysis, Training, Classification, Learning Vector Quantization, dimensionality reduction, functional data, relevance learning, supervised learning",

author = "M. Straat and M. Kaden and M. Gay and T. Villmann and Alexander Lampe and U. Seiffert and M. Biehl and F. Melchert",

year = "2017",

month = aug,

day = "31",

doi = "10.1109/WSOM.2017.8020019",

language = "English",

pages = "1--6",

booktitle = "12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)",

publisher = "IEEEXplore",

note = "12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM) ; Conference date: 28-06-2017 Through 30-06-2017",

}

Straat, M, Kaden, M, Gay, M, Villmann, T, Lampe, A, Seiffert, U, Biehl, M & Melchert, F 2017, Prototypes and matrix relevance learning in complex fourier space. in 12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM). IEEEXplore, blz. 1-6, 12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM), Nancy, France, 28/06/2017. https://doi.org/10.1109/WSOM.2017.8020019

TY - GEN

T1 - Prototypes and matrix relevance learning in complex fourier space

AU - Straat, M.

AU - Kaden, M.

AU - Gay, M.

AU - Villmann, T.

AU - Lampe, Alexander

AU - Seiffert, U.

AU - Biehl, M.

AU - Melchert, F.

PY - 2017/8/31

Y1 - 2017/8/31

N2 - In this contribution, we consider the classification of time-series and similar functional data which can be represented in complex Fourier coefficient space. We apply versions of Learning Vector Quantization (LVQ) which are suitable for complex-valued data, based on the so-called Wirtinger calculus. It makes possible the formulation of gradient based update rules in the framework of cost-function based Generalized Matrix Relevance LVQ (GMLVQ). Alternatively, we consider the concatenation of real and imaginary parts of Fourier coefficients in a real-valued feature vector and the classification of time domain representations by means of conventional GMLVQ.

AB - In this contribution, we consider the classification of time-series and similar functional data which can be represented in complex Fourier coefficient space. We apply versions of Learning Vector Quantization (LVQ) which are suitable for complex-valued data, based on the so-called Wirtinger calculus. It makes possible the formulation of gradient based update rules in the framework of cost-function based Generalized Matrix Relevance LVQ (GMLVQ). Alternatively, we consider the concatenation of real and imaginary parts of Fourier coefficients in a real-valued feature vector and the classification of time domain representations by means of conventional GMLVQ.

KW - Calculus

KW - Discrete Fourier transforms

KW - Frequency-domain analysis

KW - Prototypes

KW - Time series analysis

KW - Time-domain analysis

KW - Training

KW - Classification

KW - Learning Vector Quantization

KW - dimensionality reduction

KW - functional data

KW - relevance learning

KW - supervised learning

U2 - 10.1109/WSOM.2017.8020019

DO - 10.1109/WSOM.2017.8020019

M3 - Conference contribution

SP - 1

EP - 6

BT - 12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)

PB - IEEEXplore

T2 - 12th International Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization (WSOM)

Y2 - 28 June 2017 through 30 June 2017

ER -

Prototypes and matrix relevance learning in complex fourier space

Samenvatting

Conference

Toegang tot document

Handle.net

Activiteiten

Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization

Workshop on Self-Organizing Maps and Learning Vector Quantization, Clustering and Data Visualization

Citeer dit