DMCA
System Optimisation by Fusion of Information for Data Mining Problems
BibTeX
@MISC{Chauvin_systemoptimisation,
author = {Stéphane Chauvin and Christian L. Dunis and Jason Laws and Luis Jáñez Escalada},
title = {System Optimisation by Fusion of Information for Data Mining Problems},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
This paper presents a new fusion model called System Optimisation by Fusion of Information (SOFI) which is derived from Bayesian networks. It is designed to improve user decision-making through the help of the more active and powerful Decision Support System (DSS) that is used to reduce the “Information Gap ” for data mining applications. The SOFI module is a software which can be integrated in every kind of data mining platform that gathers several data mining tools. It is especially designed for the Knowledge Discovery in Databases (KDD). After accumulating various analyses, the user has quick access to a synthetic result and can rebuild a complete analysis. The fusion process is specified as a multi-dimensional classifier. It is introduced for the recognition of pattern classes after cumulating supervised and unsupervised learning. We review the components of data mining technology to provide the material for fusion. We focus our presentation over the logical process that is used to discover knowledge and for which the a priori information is integrated step by step. This original SOFI model of fusion automatically generates a Bayesian Network which produces a logical meaning from a data set. The Bayesian Network is optimised by a stochastic process in order to discover decision rules into data. We reserve a large section for the SOFI application which is dedicated here to produce a model classification for forecasting problems. We conclude with the model performance of the SOFI model. 1 1
Keyphrases
system optimisation data mining problem bayesian network unsupervised learning synthetic result decision rule fusion process original sofi model user decision-making priori information complete analysis quick access model classification pattern class large section logical process model performance sofi application powerful decision support system new fusion model multi-dimensional classifier various analysis data mining application sofi module gather several data mining tool stochastic process logical meaning data set knowledge discovery sofi model information gap
