Face verification based on convolutional neural network and deep learning

This paper presents the new face verification algorithm based on deep convolutional neural network. The algorithm produces face feature vectors, distance between these vectors allows to determine whether images from the same class. Comparative experimental results are given for LFW test database and modern face recognition algorithms. ROC-curve and equal error rate are used to determine the accuracy of compared algorithms. Testing was carried out under the “image restricted” verification paradigm. With unsupervised learning, the algorithm can't have any access to the data class labels, the statistics of these labels, or the means of generating these labels. Proposed face verification algorithm is more accurate than other modern algorithms.