|
Vol: 47(61) No: 2 / June 2002 Classification by Genetic Algorithm Optimization Lucian V. Boiculese Department of Medical Informatics and Biostatistics, University of Medicine and Pharmacy "Gr.T.Popa", Faculty of Medicine, 16 University Street, 6600 Iasi, Romania, phone: 0232-215350, e-mail: lboiculese@mail.com Mihaela Moscalu Department of Medical Informatics and Biostatistics, University of Medicine and Pharmacy "Gr.T.Popa", Faculty of Bioengineering, 16 University Street, 6600 Iasi, Romania, phone: 0232-215350, e-mail: mroxy@umfiasi.ro Keywords: genetic algorithm, fuzzy system, optimization, classification, neural networks. Abstract Human logic inference usually operates with imprecise information. Therefore there are points that belong to more than one class. Fuzzy systems are suitable for this application as they work with imprecision that model the belonging to a specific multitude. Training a fuzzy system is a problem of optimization. It is known that learning methods for neural networks systems are very well developed. Taking into account that a neural network can implement a fuzzy system, it is possible to profit of both fuzzy and neural network methods. A fuzzy neural network system that implement the human logic inference “If facts then conclusion” was developed. The system was tested in order to classify the etiological agents like mycobacteria. The back-propagation training method and also the genetic algorithm technique were applied in order to increase the chances of optimization. References [1] H.J. Zimmerman, Fuzzy set Theory and its Applications. Kluwer Academic Publishers 1995. [2] D. Dumitrescu, Algoritmi Genetici si Strategii Evolutive – aplicatii in Inteligenta Artificiala si in domenii conexe. Editura Albastra, Cluj Napoca 2000. [3] Jan Zizka, Learning Control Rules for Takagi-Sugeno Fuzzy Controllers using Genetic Algorithms, EUFIT’96 September 2-5 1996. [4] Teodorescu Horia-Nicolai, Fuzzy Logic and Fuzzy Systems in Medicine and Biomedical Engineering. A historical perspective, Fuzzy Systems & A.I. Reports & Letters. Romanian Academy Publishing House, Iasi Romania 1998. [5] Yun Li and Kim Chwee NG , Genetic Algorithm Based Techniques for Design Automation of Three Term Fuzzy Systems, http://www.elec.gla.ac.uk/reports/csc95008.htm, 1995 . 6] F. Wieland, F. Aliev, Neuro-Fuzzy-Genetic Adaptive Control System, Fourth European Congress on Inteligent Techniques and soft Computing, Aachen Germany, September 2-5 , 1996. [7] Li-Xin Wang, A Course in Fuzzy System and Control, Prentince-Hall International, Inc. USA 1997. |