TR
Avrasya Ekonometri �statistik ve Ampirik Ekonomi DergisiYl:2019 Say: 14 Alan: Ampirik ktisat

Muhammet ATALAY
Yapay Sinir Alar ve Bulank Mantn Birlikte Kullanld Hibrit Modeller ve Uygulamalar
 
Yapay zek tekniklerinden her birinin kendine zg yetenekleri bulunmaktadr. Yapay sinir alar insann sinir sistemini taklit ederek bilgisayar renmesini gerekletirir. Bulank mantk ise insann dn tarzna ok yakndr. Ayn zamanda szel deikenleri de kullanabilmektedir. Yalnz bu tekniklerinin kendine zg dezavantajlar da bulunmaktadr. Bulank tasarml sistemlerin en nemli dezavantaj bu sistemlerin renme yeteneinin olmamasdr. Yapay zek teknolojilerindeki ilerlemelere paralel olarak bu tekniklerin birlikte kullanm ile bu yntemlerin dezavantajlar ortadan kaldrlmaya allmtr. Bu almada, birer yapay zeka teknii olan bu iki yntemin bir arada kullanld hibrit (melez) yntem yaklamlar tartlm ve bu hibrit yntemlerin kullanlarak zellikle sosyal bilimler alannda yaplm almalardan rnekler sunulmutur. Bulank sinir alar ve sinirsel bulank a modelleri yaklamlar almann odakland yntemlerdir. Grlmektedir ki bu almalar uzun yllardr yaplagelmekte olup, verinin eitliliinin, boyutunun ve hznn artt son yllarda da farkl alanlarda iyi sonular veren modeller nerilmektedir.

Anahtar Kelimeler: : fuzzy logic, artificial neural networks, neural fuzzy systems, fuzzy neural networks, hybrid models, social sciences


Hybrid Models Using Artificial Neural Networks and Fuzzy Logic and Applications
 
Each of the artificial intelligence techniques has unique abilities. Artificial neural networks perform computer learning by imitating human nervous system. Fuzzy logic is very close to human thinking. It can also use verbal variables. However, these techniques have their own disadvantages. The most important disadvantage of fuzzy design is that these systems do not have the ability to learn. In parallel with the advances in artificial intelligence technologies, the disadvantages of these methods have been tried to be eliminated by using these techniques together. In this study, hybrid method approaches using artificial intelligence techniques are discussed. Examples of the studies conducted in the field of social sciences using these hybrid methods are presented. Fuzzy neural networks and neural fuzzy network models are the methods that the study focuses on. It is seen that these studies have been carried out for many years. In recent years, when the variety, size and speed of data have increased, models have been proposed that give good results in different areas.

Keywords: multivariate statistics, welfare economics, principal components analysis, regional development ranking, market institutionalization.


Detay

ÇERK