Convolutional Neural Network Face Recognition Method Using Fisher’s Criterion

In order to further improve the ability of automatic feature extraction deep learning technology and better solve problem system recognition declining when number training samples is reduced or iterations reduced, this paper proposes a fisher-based convolutional neural network algorithm, which automatically obtains structural information face image with help algorithm then reduces weights reduce complexity model. Through test, verification data shows that error rate can be controlled at about 10% by extracting 10 14 pictures each type for training. When images are improved.