ICANet: a simple cascade linear convolution network for face recognition

• Published in: EURASIP journal on image and video processing 2018-06, Vol.2018 (1), p.1-7, Article 51
• Main Authors: Zhang, Yongqing, Geng, Tianyu, Wu, Xi, Zhou, Jiliu, Gao, Dongrui
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Augmented reality; Biometrics; Convolution; Convolution network; Engineering; Face recognition; Facial recognition technology; ICANet; Image classification; Image Processing and Computer Vision; Independent component analysis; Machine learning; Pattern Recognition; Signal,Image and Speech Processing; Visual Information Learning and Analytics on Cross-Media Big Data
• ISSN: 1687-5281; 1687-5176; 1687-5281
• DOI: 10.1186/s13640-018-0288-4

Recently, deep convolutional networks have demonstrated their capability of improving the discriminative power compared with other machine learning method, but its feature learning mechanism is not very clear. In this paper, we present a cascaded linear convolutional network, based on independent component analysis (ICA) filters, named ICANet. ICANet consists of three parts: a convolutional layer, a binary hash, and a block histogram. It has the following advantages over other methods: (1) the network structure is simple and computationally efficient, (2) the ICA filter is trained with an unsupervised algorithm using unlabeled samples, which is practical, and (3) compared to deep learning models, each layer parameter in ICANet can be easily trained. Thus, ICANet can be used as a benchmark for the application of a deep learning framework for large-scale image classification. Finally, we test two public databases, AR and FERET, showing that ICANet performs well in facial recognition tasks.