Additive Parameter for Deep Face Recognition

作     者：Jamshaid Ul Rahman Qing Chen Zhouwang Yang 

作者机构：School of Mathematical SciencesUniversity of Science and Technology of China96 Jinzhai RoadHefei 230026AnhuiPeople’s Republic of China 

出 版 物：《Communications in Mathematics and Statistics》 (数学与统计通讯（英文）)

年 卷 期：2020年第8卷第2期

页      面：203-217页

核心收录：

学科分类：08[工学] 080203[工学-机械设计及理论] 0802[工学-机械工程] 

基　　金：The work is supported by the NSF of China(No.11871447) Anhui Initiative in Quantum Information Technologies(AHY150200) 

主　　题：Additive parameter Angular margin Deep convolutional neural networks Face recognition Softmax loss 

摘      要：The performance of feature learning for deep convolutional neural networks(DCNNs)is increasing promptly with significant improvement in numerous *** studies on loss functions clearly describing that better normalization is helpful for improving the performance of face recognition(FR).Several methods based on different loss functions have been proposed for FR to obtain discriminative *** this paper,we propose an additive parameter depending on multiplicative angular margin to improve the discriminative power of feature embedding that can be easily *** additive parameter approach,an automatic adjustment of the seedling element as the result of angular marginal seed is offered in a particular way for the angular softmax to learn angularly discriminative *** train the model on publically available dataset CASIA-WebFace,and our experiments on famous benchmarks YouTube Faces(YTF)and labeled face in the wild(LFW)achieve better performance than the various state-of-the-art approaches.