Preparing for knowledge extraction in modular neural networks

Lambert Spaanenburg; Cornelis H. Slump; Rienk Venema; B.J. van der Zwaag

Preparing for knowledge extraction in modular neural networks

Lambert Spaanenburg
, Cornelis H. Slump
, Rienk Venema
, B.J. van der Zwaag

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

69 Downloads (Pure)

Abstract

Neural networks learn knowledge from data. For a monolithic structure, this knowledge can be easily used but not isolated. The many degrees of freedom while learning make knowledge extraction a computationally intensive process as the representation is not unique. Where existing knowledge is inserted to initialize the network for training, the effect becomes subsequently randomized within the solution space. The paper describes structuring techniques such as modularity and hierarchy to create a topology that provides a better view on the learned knowledge to support a later rule extraction.

Original language	Undefined
Title of host publication	3rd IEEE Benelux Signal Processing Symposium (SPS-2002)
Place of Publication	Leuven, Belgium
Publisher	Katholieke Universiteit
Pages	121-124
Number of pages	4
ISBN (Print)	-
Publication status	Published - Mar 2002
Event	3rd IEEE Benelux Signal Processing Symposium, SPS-2002 - Leuven, Belgium Duration: 21 Mar 2002 → 22 Mar 2002 Conference number: 3

Conference

Conference	3rd IEEE Benelux Signal Processing Symposium, SPS-2002
Abbreviated title	SPS
Country/Territory	Belgium
City	Leuven
Period	21/03/02 → 22/03/02

Keywords

EWI-1438
IR-43136
METIS-205958

Access to Document

Cite this

@inproceedings{5d5a8a9586454594a21d3ec67951d46f,

title = "Preparing for knowledge extraction in modular neural networks",

abstract = "Neural networks learn knowledge from data. For a monolithic structure, this knowledge can be easily used but not isolated. The many degrees of freedom while learning make knowledge extraction a computationally intensive process as the representation is not unique. Where existing knowledge is inserted to initialize the network for training, the effect becomes subsequently randomized within the solution space. The paper describes structuring techniques such as modularity and hierarchy to create a topology that provides a better view on the learned knowledge to support a later rule extraction.",

keywords = "EWI-1438, IR-43136, METIS-205958",

author = "Lambert Spaanenburg and Slump, \{Cornelis H.\} and Rienk Venema and \{van der Zwaag\}, B.J.",

note = "SAS 023N02 ; 3rd IEEE Benelux Signal Processing Symposium, SPS-2002 ; Conference date: 21-03-2002 Through 22-03-2002",

year = "2002",

month = mar,

language = "Undefined",

isbn = "-",

pages = "121--124",

booktitle = "3rd IEEE Benelux Signal Processing Symposium (SPS-2002)",

publisher = "Katholieke Universiteit",

}

TY - GEN

T1 - Preparing for knowledge extraction in modular neural networks

AU - Spaanenburg, Lambert

AU - Slump, Cornelis H.

AU - Venema, Rienk

AU - van der Zwaag, B.J.

N1 - Conference code: 3

PY - 2002/3

Y1 - 2002/3

N2 - Neural networks learn knowledge from data. For a monolithic structure, this knowledge can be easily used but not isolated. The many degrees of freedom while learning make knowledge extraction a computationally intensive process as the representation is not unique. Where existing knowledge is inserted to initialize the network for training, the effect becomes subsequently randomized within the solution space. The paper describes structuring techniques such as modularity and hierarchy to create a topology that provides a better view on the learned knowledge to support a later rule extraction.

AB - Neural networks learn knowledge from data. For a monolithic structure, this knowledge can be easily used but not isolated. The many degrees of freedom while learning make knowledge extraction a computationally intensive process as the representation is not unique. Where existing knowledge is inserted to initialize the network for training, the effect becomes subsequently randomized within the solution space. The paper describes structuring techniques such as modularity and hierarchy to create a topology that provides a better view on the learned knowledge to support a later rule extraction.

KW - EWI-1438

KW - IR-43136

KW - METIS-205958

M3 - Conference contribution

SN - -

SP - 121

EP - 124

BT - 3rd IEEE Benelux Signal Processing Symposium (SPS-2002)

PB - Katholieke Universiteit

CY - Leuven, Belgium

T2 - 3rd IEEE Benelux Signal Processing Symposium, SPS-2002

Y2 - 21 March 2002 through 22 March 2002

ER -