Unconstrained face identification using maximum likelihood of distances between deep off-the-shelf features
作者：
Highlights：
• Maximum likelihoods estimate of distances increases face recognition accuracy.
• Likelihood is estimated using non-central chi-squared distribution of distances.
• Regularization enforces similarity of any instance and input image to be close to similarity of this instance and decision class.
摘要
•Maximum likelihoods estimate of distances increases face recognition accuracy.•Likelihood is estimated using non-central chi-squared distribution of distances.•Regularization enforces similarity of any instance and input image to be close to similarity of this instance and decision class.
论文关键词：Statistical pattern recognition,Unconstrained face recognition,Maximum likelihood estimation,CNN (Convolution neural network),Kullback–Leibler divergence,Off-the-shelf deep features
论文评审过程：Received 13 October 2017, Revised 2 April 2018, Accepted 29 April 2018, Available online 9 May 2018, Version of Record 17 May 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.04.039
