Kıvanç KILIÇ, Deniz BAŞ, İsmail Hakkı BOYACI

Optimum Yapay Sınır Ağı Yapısının Belirlenmesinde Kolay bir Yaklaşım (İngilizce)

Yüksek performansa sahip sinir agının optimum topolojisinin ve parametrelerinin belirlenmesi emek-yogun bir islemdir. Bu çalısma kapsamında, optimum ag yapısının bir kod yardımıyla belirlenmesi incelenmistir. Bu dogrultuda, MATLAB® ile olusturulan kod kullanılarak incelenen parametre aralıgındaki sinir aglarının olusturulması ve egitimi gerçeklestirilmistir. Ayrıca, kullanılan kod yardımıyla elde edilen aglar test edilmekte ve seçilen performans parametrelerinin (Mutlak Ortalama Sapma, Ortalama Karesel Hata ve Determinasyon Katsayısı) hesaplanmaktadır. Çalısma kapsamında, geri yayılımlı ileri beslemeli çok katmanlı sinir agı yapısı model olarak seçilmistir. Seçilen ag yapısı için incelenen parametreler; sinir hücresi sayısı, egitim fonksiyonu ve transfer fonksiyonu olarak belirlenmistir. Bunlara ek olarak, olusturulan kod üç farklı veri seti kullanılarak test edilmistir. Elde edilen sonuçlar, olusturulan kodun kinetik çalısmalar sonucunda elde edilen verilerin incelenmesinde kullanılacak ag topolojisi ve parametrelerinin belirlenmesinde basarılı bir sekilde uygulanacagını göstermektedir.

Anahtar Kelimeler:

Yapay sinir agları, kinetik uygulama, optimizasyon, modelleme

An Easy Approach for the Selection of Optimal Neural Network Structure (In English)

Investigation of an optimal topology and parameters of a high performing neural network is a labourintensive process. In this study, the possibility of obtaining the optimal network structure using a script was investigated. For this purpose, making use of a script coded in MATLAB®, all possible networks in the range of investigated network parameters were created and trained. In addition, the networks created with the help of the script were tested and the selected performance parameters (Absolute Average Deviation, Mean Squared Error, and Coefficient of Determination) were calculated. e network model studied was chosen as multilayer perceptron with a feed-forward back propagation algorithm. e investigated parameters of the selected network were number of neurons, training functions, and transfer functions. Additionally, the script was tested on three different data sets. e results indicate that the script can be successfully applied for identification of the network topology and the parameters to be used for the evaluation of the data obtained from kinetic researches.

Keywords:

Artificial neural networks, kinetic application, optimization, modeling ,

PDF

___

Fausett LV. 1994. Fundamentals of neural networks: architectures, algorithms and applications, Prentice Hall, 461 p, New Jersey.
Kılıç K, Boyacı İH, Köksel H, Küsmenoğlu İ. 2007. A classification system for beans using computer vision system and artificial neural networks, J. Food Eng., 78: 897–904.
Baş D, Boyacı İH. 2007. Modeling and optimization II: Comparison of estimation capabilities of response surface methodology with artificial neural networks in a biochemical reaction, J. Food Eng., 78: 846-854.
Torrecilla JS, Otero L, Sanz PD. 2004. A neural network approach for thermal/pressure food processing, J. Food Eng. 62: 89–95.
Lou W, Nakai S. 2001. Application of artificial neural networks for predicting the thermal inactivation of bacteria: a combined effect of temperature, pH and water activity, Food Res Int., 34: 573–579.
Mittal GS, Zhang J. 2003. Artificial neural network- based psychrometric predictor, Biosystems Engineering, 85: 283–289.
Buciński A, Zieliński H, Kozłowska H. 2004. Artificial neural networks for prediction of antioxidant capacity of cruciferous sprouts, Trends Food Sci. Technol., 15: 161–169.
Razavi SMA, Mousavi SM, Mortazavi SA. 2003. Dynamic prediction of milk ultrafiltration performance: A neural network approach, Chem. Eng. Sci. 58: 4185 - 4195.
Hajmeer MN, Basheer IA, Najjar YM. 1997. Computational neural networks for predictive microbiology II. Application to microbial growth, Int. J. Food Microbiol., 34: 51-66.
Brodnjak-Vončina D, Kodba ZC, Novič M. 2005. Multivariate data analysis in classification of vegetable oils characterized by the content of fatty acids, Chemom. Intell. Lab. Syst., 75: 31– 43.
Torrecilla JS, Otero L, Sanz PD. 2005. Artificial neural networks: a promising tool to design and optimize high- pressure food processes, J. Food Eng. 69: 299–306.
Dutta R, Hines EL, Gardner JW, Udrea DD, Boilot P. 2003. Non-destructive egg freshness determination: an electronic nose based approach, Meas. Sci. Technol., 14: 190–198.
Jimenez-Marquez SA, Thibault J, Lacroix C. 2005. Prediction of moisture in cheese of commercial production using neural networks, Int. Dairy J., 15: 1156–1174.
Gonçalves EC, Minim LA, Coimbra JSR, Minim VPR. 2005. Modeling sterilization process of canned foods using artificial neural networks. Chem. Eng. Process., 44: 1269–1276.
Baş D, Dudak FC, Boyacı İH. 2007. Modeling and optimization III: Reaction rate estimation using artificial neural network (ANN) without a kinetic model, J. Food Eng., 79: 622-628. Eng., 79: 622-628. Eng., 79: 622-628.
Baş D, Dudak FC, Boyacı İH. 2007. Modeling and optimization IV: Investigation of reaction kinetics and kinetic constants using a program in which artificial neural network (ANNs) was integrated, J. Food Eng., 79: 1152-1158.