双约束的LossFunction摘录https:zhuanlan.zhihu.comp34404607SunY,ChenY,WangX,etal.Deeplear 双约束的Loss Function 摘录https://zhuanlan.zhihu.com/p/34404607 Sun Y, Chen Y, Wang X, et al. Deep learning face representation by joint iden
- 双约束的Loss Function 摘录https://zhuanlan.zhihu.com/p/34404607
Sun Y, Chen Y, Wang X, et al. Deep learning face representation by joint identification-verification [C]// NIPS, 2014. Sun Y, Wang X, Tang X. Deeply learned face representations are sparse, selective, and robust [C]// CVPR, 2015. Sun Y, Liang D, Wang X, et al. Deepid3: Face recognition with very deep neural networks [J]. arXiv, 2015. 
DeepID2, DeepID2, DeepID3都在采用Softmax Contrastive LossContrast Loss是 
同类特征的L2距离尽可能小不同类特征的L2距离大于margin(间隔) mContrastive Loss同时约束类内紧凑和类间分离。25个patch训练25个CNN特征联合后PAC降维训练Joint Bayesian分类在LFW上Softmax Contrast Loss的DeepID2达到99.15%多层加监督信息的DeepID2达到99.47采用更大的deep CNN的DeepID3达到99.53%。DeepID系列是早期的深度学习人脸识别方法但代码都没有开源而且深度特征是多patch联合还要训练分类器繁琐不实用。