Özet:

This study suggests a method using the Evolutionary Neural Network (ESA), which reduces the classification time of the image, which can keep the classification performance to acceptable value. Separate Wave Conversion - In the hybrid model called the Evolutionary Neural Network (ADD-TDA-ESA) using the ADD-TDA-based algorithmic summary function, the classification time was reduced along with ESA. The most important feature of perceptive summary functions is to obtain the clear characteristics of images. In this method, the ADD-TDA algorithmic summary function was first applied to obtain the distinctive characteristics of images. Images of the size of 32x32 consisting of distinct characteristics were submitted to the ESA and submitted to the Support Vector Machine for classification. The ADD-TDA-ESA method is applied to images in the Caltech-101 database. Experimental results showed that the recommended ADD-TDA-ESA method has a 95.8% accuracy rate. The working time, which is 241.21 seconds in the classical method, was reduced to 83.08 seconds. Test results show that the ADD-TDA-ESA method has a much faster performance than the classic ESA, keeping the image classification accuracy high.

Anahtar Kelimeler: