Face Recognition with Backpropagation Artificial Neural Networks Using the Sigmoid Activation Function

Abstract

In the current development of information technology facilities that are so rapid and complex, the reliability of a system that processes data properly will produce good information. The use of computers today is not only to help human work, but has also replaced human work in various fields. The purpose of this research is to design an image model in an application program. This existing digital image can then be used for various purposes. One way is to take photos on a driver's license (SIM), when taking photos directly using a digital camera and facial photo files are also directly stored in the police database. So this facial photo recognition program can actually also be applied by the police to recognize the faces of criminals. From the 10 experimental samples, it can be concluded that the best LearnRates with the lowest ErrorLevel for BMP format files are for learnrate 0.2 with an error rate of 18,54229 and a match rate of 81,45771 with a recognition rate of more than 60% in photo 2, while with an error rate of 51.00543 and a match level 48,99457

Analisis Fungi Aktivasi Sigmoid Biner dan Bipolar Pada Backpropagation Dalam Mengklasifikasikan Pasien Kanker Pasca Operasi

“…Perbedaan pada setiap hasil pembelajaran juga akan terjadi jika menggunakan fungsi aktivasi yang berbeda. Perbedaan instrument dalam melakukan proses klasifikasi pada backpropagation mempengaruhi proses dan hasil yang didapatkan dalam melakukan klasifikasi (Betaubun et al, 2022). Untuk mengetahui fungsi aktivasi yang terbaik dalam melakukan klasifikasi pada pembelajaran Backpropagation maka dilakukan analisis terhadap fungsi aktivasi tersebut.…”