Stein损失下BP神经网络分类方法在人脸识别中的应用

The BP neural network classification method under Stein loss function and the application to face recognition

在统计判决理论的框架下,针对一类特定目标人脸识别中存在的问题,提出了基于Stein损失的BP神经网络分类方法,证明了Stein损失下的BP神经网络的收敛性,经过剑桥大学ORL人脸库的图像识别实验,表明这种方法能有效解决传统的BP神经网络特定目标人脸识别中存在的问题.

Within the statistical decision theory framework,this paper focuses on specific objectives in face recognition,and proposes the BP neural network classification methods which is under the Stein loss function,and proves that the convergence of BP neural network under the Stein loss function.The image recognition experiments with the ORL face database in Cambridge show that this method can effectively solve the traditional problems in specific target face recognition of BP neural network.