可用于人脸识别的反馈型二元神经网络

提出和分析了一种新型的反馈型随机神经网络,并将其用于解决复杂的人脸识别问题.该模型采用随机型加权联接,神经元为简单的非线性处理单元.理论分析揭示该网络模型存在唯一的收敛平稳概率分布,当网络中神经元个数较多时,平稳概率分布逼近于Boltzmann-Gibbs分布,网络模型与马尔可夫随机场之间存在密切关系.在设计了一种新型模拟退火和渐进式Boltzmann学习算法后,系统被成功地应用于难度较大的静态和动态人像识别,实验结果证实了系统的可行性和高效率.

A Novel Recurrent Neural Network for Face Recognition

A novel stochastic neural network is proposed in this paper. Unlike the traditional Boltzmann machine, the new model uses stochastic connections rather than stochastic activation functions. Each neuron has very simple functionality but all of its synapses are stochastic. It is shown that the stationary distribution of the network uniquely exists and it is approximately a Boltzmann Gibbs distribution. It is also revealed there exists a strong relationship between the model and the Markov random field. New efficient techniques are developed to implement simulated annealing and Boltzmann lerning.The model has been successfully applied to large-scale face recognition task in which face images are dynamically captured from a video source.Learing and recoginiz-ing processes are carried out in real time.The experimental results show the new model is not only feasible but also efficient.