Identyfikatory
Warianty tytułu
Języki publikacji
Abstrakty
The main aim of Self Optimizing Neural Network (SONN), which are presented in this paper, is construction of expert system on the basis of analysis of medical information about group of patients. The expert system is built on the basis of neural network, and the main task of this system is to expect future patient health, based on information about the patient. Such a system can give the doctors a hint about that what can be happen with patient. And what is more important - the SONN construction process is very flexibly and adapts topology and all weights to training data. This is undoubtedly a great advantage of this type of neural network. Moreover the construction process is quite simple. The network topology and all connections between neurons can be easy implemented and kept in such a way, which allows to create very efficient expert system. In this paper we describe the process of construction of neural network which is based on one-shot analysis of learning patterns. On the basis of appropriate computation the SONN topology is built. The construction process can be repeated on the learning group of patients. In this way the expert system (based on SONN) will be better and better.
Rocznik
Tom
Strony
77--82
Opis fizyczny
Bibliogr. 3 poz., rys., tab.
Twórcy
autor
- University of Łódź, Faculty of Mathematics and Informatics, ul. Banacha 22, 90-238 Łódź, Poland
autor
autor
Bibliografia
- [1] S. WOJCZYK, A. POPA, M. LIZIS, Idea Samooptymalizujacych Sieci Neuronowych SONN budowanych na bazie metrycznej przestrzeni cech wejsciowych, rozdział 29 w monografii: Bazy danych- nowe technologie, Wydawnictwo Komunikacji i Łacznosci, Warszawa 2007, 315-324.
- [2] S. WOJCZYK, A. POPA, M. LIZIS, Samooptymalizujace siatkowe sieci neuronowe, rozdział 30 w monografii: Bazy danych - Nowe technologie, Wydawnictwo Komunikacji i Łacznosci, Warszawa 2007, 325- 333.
- [3] A. HORZYK, R. TADEUSIEWICZ, Self-Optimizing Neural Networks, Advances in Neural Networks - ISNN 2004, Yin F., Wang J., Guo C. (eds), (Springer-Verlag, Berlin Heidelberg, 2004) 150-155.
Typ dokumentu
Bibliografia
Identyfikator YADDA
bwmeta1.element.baztech-article-PWA4-0006-0010