Pengaruh Faktor Proporsional Pada Jaringan Saraf Propagasi Balik Untuk Pengenalan Wajah Berbasis Eigenfaces
Quarthano Reavindo · 2010
At this time the standard backpropagation has been developed by 2 learning factors, that are learning rate (α) and momentum (β). The third factor that is called proportional factor (γ) is used in this research to control its influence for the recognition of the face that based on eigenfaces. The influence of this factor can be measured by passing through the comparation performance of the two kinds backpropagation above. The network performance that are used as the basics are the convergence speed of the network, the ability of network memorization and the network generalization. By passing through the measurement of the performances so this research concludes that proportional factor will bring bad performance about the neural network when it is used at the interval [0.1 , 0.9] and [0.01 , 0.09] but at the interval [0.001 , 0.009] it will give good performance. vi Universitas Sumatera Utara