Portable Rule Extraction Method for Neural Network Decisions Reasoning
Neural network (NN) methods are sometimes useless in practical applications, because they are not properly tailored to the particular market's needs. We focus thereinafter specifically on financial market applications. NNs have not gained full acceptance here yet. One of the main reasons is the...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
International Institute of Informatics and Cybernetics
2005-08-01
|
Series: | Journal of Systemics, Cybernetics and Informatics |
Subjects: | |
Online Access: | http://www.iiisci.org/Journal/CV$/sci/pdfs/P185508.pdf
|
id |
doaj-3f95bfe5241241c0a36c23b1033be60c |
---|---|
record_format |
Article |
spelling |
doaj-3f95bfe5241241c0a36c23b1033be60c2020-11-24T22:49:39ZengInternational Institute of Informatics and CyberneticsJournal of Systemics, Cybernetics and Informatics1690-45242005-08-01347984Portable Rule Extraction Method for Neural Network Decisions ReasoningDarius PLIKYNAS0Leonas SIMANAUSKAS1Ausra Rasteniene2 IT Department, Vilnius Management Academy IT Department, Vilnius Management Academy Management Department, Vilnius Management Academy Neural network (NN) methods are sometimes useless in practical applications, because they are not properly tailored to the particular market's needs. We focus thereinafter specifically on financial market applications. NNs have not gained full acceptance here yet. One of the main reasons is the "Black Box" problem (lack of the NN decisions explanatory power). There are though some NN decisions rule extraction methods like decompositional, pedagogical or eclectic, but they suffer from low portability of the rule extraction technique across various neural net architectures, high level of granularity, algorithmic sophistication of the rule extraction technique etc. The authors propose to eliminate some known drawbacks using an innovative extension of the pedagogical approach. The idea is exposed by the use of a widespread MLP neural net (as a common tool in the financial problems' domain) and SOM (input data space clusterization). The feedback of both nets' performance is related and targeted through the iteration cycle by achievement of the best matching between the decision space fragments and input data space clusters. Three sets of rules are generated algorithmically or by fuzzy membership functions. Empirical validation of the common financial benchmark problems is conducted with an appropriately prepared software solution.http://www.iiisci.org/Journal/CV$/sci/pdfs/P185508.pdf Data MiningNeural networksInformation ExtractionFuzzy LogicDecisions Reasoning |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Darius PLIKYNAS Leonas SIMANAUSKAS Ausra Rasteniene |
spellingShingle |
Darius PLIKYNAS Leonas SIMANAUSKAS Ausra Rasteniene Portable Rule Extraction Method for Neural Network Decisions Reasoning Journal of Systemics, Cybernetics and Informatics Data Mining Neural networks Information Extraction Fuzzy Logic Decisions Reasoning |
author_facet |
Darius PLIKYNAS Leonas SIMANAUSKAS Ausra Rasteniene |
author_sort |
Darius PLIKYNAS |
title |
Portable Rule Extraction Method for Neural Network Decisions Reasoning |
title_short |
Portable Rule Extraction Method for Neural Network Decisions Reasoning |
title_full |
Portable Rule Extraction Method for Neural Network Decisions Reasoning |
title_fullStr |
Portable Rule Extraction Method for Neural Network Decisions Reasoning |
title_full_unstemmed |
Portable Rule Extraction Method for Neural Network Decisions Reasoning |
title_sort |
portable rule extraction method for neural network decisions reasoning |
publisher |
International Institute of Informatics and Cybernetics |
series |
Journal of Systemics, Cybernetics and Informatics |
issn |
1690-4524 |
publishDate |
2005-08-01 |
description |
Neural network (NN) methods are sometimes useless in practical applications, because they are not properly tailored to the particular market's needs. We focus thereinafter specifically on financial market applications. NNs have not gained full acceptance here yet. One of the main reasons is the "Black Box" problem (lack of the NN decisions explanatory power). There are though some NN decisions rule extraction methods like decompositional, pedagogical or eclectic, but they suffer from low portability of the rule extraction technique across various neural net architectures, high level of granularity, algorithmic sophistication of the rule extraction technique etc. The authors propose to eliminate some known drawbacks using an innovative extension of the pedagogical approach. The idea is exposed by the use of a widespread MLP neural net (as a common tool in the financial problems' domain) and SOM (input data space clusterization). The feedback of both nets' performance is related and targeted through the iteration cycle by achievement of the best matching between the decision space fragments and input data space clusters. Three sets of rules are generated algorithmically or by fuzzy membership functions. Empirical validation of the common financial benchmark problems is conducted with an appropriately prepared software solution. |
topic |
Data Mining Neural networks Information Extraction Fuzzy Logic Decisions Reasoning |
url |
http://www.iiisci.org/Journal/CV$/sci/pdfs/P185508.pdf
|
work_keys_str_mv |
AT dariusplikynas portableruleextractionmethodforneuralnetworkdecisionsreasoning AT leonassimanauskas portableruleextractionmethodforneuralnetworkdecisionsreasoning AT ausrarasteniene portableruleextractionmethodforneuralnetworkdecisionsreasoning |
_version_ |
1725675429543542784 |