Classification of Raisin Grains Using Different Artificial Neural Network Methods

In addition to its nutritional properties, raisins are also a beneficial food in terms of health due to its vitamins, minerals, antioxidants and phenolic compounds. Turkey ranks first in global raisin production with a production capacity of 24%. Many problems are encountered in the classification of raisins according to their type and quality by traditional methods. In order to overcome these problems, artificial intelligence systems, whose usage area is increasing day by day, are utilized. In this study, raisin grains were classified using 3 different Artificial Neural Network (ANN) methods using the ‘Raisin’ dataset from the UCI Machine Learning Repository. Performance measurements of Competitive Layer Neural Network (CLNN), Pattern Recognition Artificial Neural Network (PRNN) and Self-Organizing Map (SOM) methods used in classification were performed. In the obtained performance measurements, PRNN has the highest success, while SOM is weaker compared to the other two methods. CLNN, on the other hand, remains at similar levels to PRNN and offers a good alternative to PRNN.