Symbolic knowledge extraction from trained neural networks: A sound approach

A. S. D'Avila Garcez; K. Broda; D. M. Gabbay

doi:10.1016/S0004-3702(00)00077-1

Symbolic knowledge extraction from trained neural networks: A sound approach

A. S. D'Avila Garcez, K. Broda, D. M. Gabbay

Research output: Contribution to journal › Article › peer-review

177 Scopus citations

Abstract

Some of the main problems of knowledge extraction methods were discussed. The results of an empirical analysis of the extraction system, using traditional examples and real-world application problems were presented. The results have shown that a very high fidelity between the extracted set of rules and the network can be achieved.

Original language	English
Pages (from-to)	155-207
Number of pages	53
Journal	Artificial Intelligence
Volume	125
Issue number	1-2
DOIs	https://doi.org/10.1016/S0004-3702(00)00077-1
State	Published - Jan 2001
Externally published	Yes

Bibliographical note

Funding Information:
We are grateful to Gerson Zaverucha, Valmir Barbosa, Luis Alfredo Carvalho, Steffen Hoelldobler, Franz Kurfess and Luis Lamb for useful discussions. We would especially like to thank Alberto de Souza, Stefan Rueger and the anonymous referees for their comments. The first author was partially supported by the Brazilian Research Agency CAPES.

Funding

We are grateful to Gerson Zaverucha, Valmir Barbosa, Luis Alfredo Carvalho, Steffen Hoelldobler, Franz Kurfess and Luis Lamb for useful discussions. We would especially like to thank Alberto de Souza, Stefan Rueger and the anonymous referees for their comments. The first author was partially supported by the Brazilian Research Agency CAPES.

Funders	Funder number
Brazilian Research Agency CAPES

Access to Document

10.1016/S0004-3702(00)00077-1

Cite this

@article{052140e3ce2549cf8d153e597f3d76e1,

title = "Symbolic knowledge extraction from trained neural networks: A sound approach",

abstract = "Some of the main problems of knowledge extraction methods were discussed. The results of an empirical analysis of the extraction system, using traditional examples and real-world application problems were presented. The results have shown that a very high fidelity between the extracted set of rules and the network can be achieved.",

author = "{D'Avila Garcez}, {A. S.} and K. Broda and Gabbay, {D. M.}",

note = "Funding Information: We are grateful to Gerson Zaverucha, Valmir Barbosa, Luis Alfredo Carvalho, Steffen Hoelldobler, Franz Kurfess and Luis Lamb for useful discussions. We would especially like to thank Alberto de Souza, Stefan Rueger and the anonymous referees for their comments. The first author was partially supported by the Brazilian Research Agency CAPES.",

year = "2001",

month = jan,

doi = "10.1016/S0004-3702(00)00077-1",

language = "אנגלית",

volume = "125",

pages = "155--207",

journal = "Artificial Intelligence",

issn = "0004-3702",

publisher = "Elsevier B.V.",

number = "1-2",

}

TY - JOUR

T1 - Symbolic knowledge extraction from trained neural networks

T2 - A sound approach

AU - D'Avila Garcez, A. S.

AU - Broda, K.

AU - Gabbay, D. M.

N1 - Funding Information: We are grateful to Gerson Zaverucha, Valmir Barbosa, Luis Alfredo Carvalho, Steffen Hoelldobler, Franz Kurfess and Luis Lamb for useful discussions. We would especially like to thank Alberto de Souza, Stefan Rueger and the anonymous referees for their comments. The first author was partially supported by the Brazilian Research Agency CAPES.

PY - 2001/1

Y1 - 2001/1

N2 - Some of the main problems of knowledge extraction methods were discussed. The results of an empirical analysis of the extraction system, using traditional examples and real-world application problems were presented. The results have shown that a very high fidelity between the extracted set of rules and the network can be achieved.

AB - Some of the main problems of knowledge extraction methods were discussed. The results of an empirical analysis of the extraction system, using traditional examples and real-world application problems were presented. The results have shown that a very high fidelity between the extracted set of rules and the network can be achieved.

UR - http://www.scopus.com/inward/record.url?scp=0035127989&partnerID=8YFLogxK

U2 - 10.1016/S0004-3702(00)00077-1

DO - 10.1016/S0004-3702(00)00077-1

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:0035127989

SN - 0004-3702

VL - 125

SP - 155

EP - 207

JO - Artificial Intelligence

JF - Artificial Intelligence

IS - 1-2

ER -

Symbolic knowledge extraction from trained neural networks: A sound approach

Abstract

Bibliographical note

Funding

Access to Document

Other files and links

Fingerprint

Cite this