Abstract: Aiming at the problem of overfitting, slow convergence, local optimization and low recognition accuracy of existing Convolutional Neural Network（CNN） in face recognition training, a new LeNet-FC convolutional neural network model is proposed in this paper.By improving the network structure of the network layer, reducing the convolution kernel, and using the optimized Logarithmic Rectified Linear Unit（L_ReLU） activation function, the recognition accuracy of the model in face recognition training reaches 99.85%.Simultaneously, a face recognition system is designed based on the LeNet-FC convolutional neural network model. The recognition accuracy of the system in the simulation test experiment with ORL face database reaches 96%.

Key words: artificial intelligence, face recognition, convolutional neural network, network structure improvement, activation function optimization

摘要： 针对已有的卷积神经网络（Convolutional Neural Network，CNN）在人脸识别训练中出现过拟合、收敛速度慢以及识别准确率不高的问题，提出了新型的LeNet-FC卷积神经网络模型。通过增加网络层、缩小卷积核等结构改进以及采用优化的对数—修正线性单元（Logarithmic Rectified Linear Unit，L_ReLU）激活函数，该模型在人脸识别训练的准确率达到了99.85%。同时基于LeNet-FC卷积神经网络模型设计了一个人脸识别系统。该系统在ORL人脸库的仿真测试实验中识别准确率达到了96%。

关键词: 人工智能, 人脸识别, 卷积神经网络, 结构改进, 激活函数优化