Sparsity issues in self-organizing-maps for structures

Markus Hagenbuchner, Giovanni Martino, Ah Chung Tsoi, Alessandro Sperduti

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

2 Citations (Scopus)

Abstract

Recent developments with Self-Organizing Maps (SOMs) produced methods capable of clustering graph structured data onto a fixed dimensional display space. These methods have been applied successfully to a number of benchmark problems and produced state-of-the-art results. This paper discusses a limitation of the most powerful version of these SOMs, known as probability measure graph SOMs (PMGraphSOMs), viz., the sparsity induced by processing a large number of small graphs, which prevents a successful application of PMGraphSOM to such problems. An approach using the idea of compactifying the generated state space to address this sparsity problem is proposed. An application to an established benchmark problem, viz., the Mutag dataset in toxicology will show that the proposed method is effective when dealing with a large number of small graphs. Hence, this work fills a gap between the processing of a number of small graphs, and the processing of densely connected graphs using PMGraphSOMs.

Original languageEnglish
Title of host publicationESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Pages35-40
Number of pages6
Publication statusPublished - 2010
Externally publishedYes

Fingerprint

Self organizing maps
Processing
Display devices

ASJC Scopus subject areas

  • Artificial Intelligence
  • Information Systems

Cite this

Hagenbuchner, M., Martino, G., Tsoi, A. C., & Sperduti, A. (2010). Sparsity issues in self-organizing-maps for structures. In ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (pp. 35-40)

Sparsity issues in self-organizing-maps for structures. / Hagenbuchner, Markus; Martino, Giovanni; Tsoi, Ah Chung; Sperduti, Alessandro.

ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. 2010. p. 35-40.

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

Hagenbuchner, M, Martino, G, Tsoi, AC & Sperduti, A 2010, Sparsity issues in self-organizing-maps for structures. in ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. pp. 35-40.
Hagenbuchner M, Martino G, Tsoi AC, Sperduti A. Sparsity issues in self-organizing-maps for structures. In ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. 2010. p. 35-40
Hagenbuchner, Markus ; Martino, Giovanni ; Tsoi, Ah Chung ; Sperduti, Alessandro. / Sparsity issues in self-organizing-maps for structures. ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. 2010. pp. 35-40
@inproceedings{c7c27f0f1cf748689ab612af69aa3260,
title = "Sparsity issues in self-organizing-maps for structures",
abstract = "Recent developments with Self-Organizing Maps (SOMs) produced methods capable of clustering graph structured data onto a fixed dimensional display space. These methods have been applied successfully to a number of benchmark problems and produced state-of-the-art results. This paper discusses a limitation of the most powerful version of these SOMs, known as probability measure graph SOMs (PMGraphSOMs), viz., the sparsity induced by processing a large number of small graphs, which prevents a successful application of PMGraphSOM to such problems. An approach using the idea of compactifying the generated state space to address this sparsity problem is proposed. An application to an established benchmark problem, viz., the Mutag dataset in toxicology will show that the proposed method is effective when dealing with a large number of small graphs. Hence, this work fills a gap between the processing of a number of small graphs, and the processing of densely connected graphs using PMGraphSOMs.",
author = "Markus Hagenbuchner and Giovanni Martino and Tsoi, {Ah Chung} and Alessandro Sperduti",
year = "2010",
language = "English",
isbn = "9782874190445",
pages = "35--40",
booktitle = "ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning",

}

TY - GEN

T1 - Sparsity issues in self-organizing-maps for structures

AU - Hagenbuchner, Markus

AU - Martino, Giovanni

AU - Tsoi, Ah Chung

AU - Sperduti, Alessandro

PY - 2010

Y1 - 2010

N2 - Recent developments with Self-Organizing Maps (SOMs) produced methods capable of clustering graph structured data onto a fixed dimensional display space. These methods have been applied successfully to a number of benchmark problems and produced state-of-the-art results. This paper discusses a limitation of the most powerful version of these SOMs, known as probability measure graph SOMs (PMGraphSOMs), viz., the sparsity induced by processing a large number of small graphs, which prevents a successful application of PMGraphSOM to such problems. An approach using the idea of compactifying the generated state space to address this sparsity problem is proposed. An application to an established benchmark problem, viz., the Mutag dataset in toxicology will show that the proposed method is effective when dealing with a large number of small graphs. Hence, this work fills a gap between the processing of a number of small graphs, and the processing of densely connected graphs using PMGraphSOMs.

AB - Recent developments with Self-Organizing Maps (SOMs) produced methods capable of clustering graph structured data onto a fixed dimensional display space. These methods have been applied successfully to a number of benchmark problems and produced state-of-the-art results. This paper discusses a limitation of the most powerful version of these SOMs, known as probability measure graph SOMs (PMGraphSOMs), viz., the sparsity induced by processing a large number of small graphs, which prevents a successful application of PMGraphSOM to such problems. An approach using the idea of compactifying the generated state space to address this sparsity problem is proposed. An application to an established benchmark problem, viz., the Mutag dataset in toxicology will show that the proposed method is effective when dealing with a large number of small graphs. Hence, this work fills a gap between the processing of a number of small graphs, and the processing of densely connected graphs using PMGraphSOMs.

UR - http://www.scopus.com/inward/record.url?scp=84887096496&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887096496&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84887096496

SN - 9782874190445

SP - 35

EP - 40

BT - ESANN 2011 proceedings, 19th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning

ER -