LOCAL BINARY PATTERN NETWORK : A DEEP LEARNING APPROACH FOR FACE RECOGNITION

Deep learning is well known as a method to extract hierarchical representations of data. In this paper a novel unsupervised deep learning based methodology, named Local Binary Pattern Network (LBPNet), is proposed to efficiently extract and compare high-level over-complete features in multilayer hierarchy. The LBPNet retains the same topology of Convolutional Neural Network (CNN) - one of the most well studied deep learning architectures - whereas the trainable kernels are replaced by the off-the-shelf computer vision descriptor (i.e., LBP). This enables the LBPNet to achieve a high recognition accuracy without requiring any costly model learning approach on massive data. Through extensive numerical experiments using the public benchmarks (i.e., FERET and LFW), LBPNet has shown that it is comparable to other unsupervised methods.