弹性RBF神经网络在人脸识别中的应用研究
|
摘要
为了提高RBF神经网络在人脸识别中的性能,应用弹性RBF神经网络设计了一种新的人脸识别方法。该方法基于神经元的活跃度对神经元进行分裂,根据神经元修复准则调整神经元之间的连接关系,从而改变RBF神经网络隐含层神经元结构以及隐含层与输出成神经元之间的连接权值,获得一种弹性RBF神经网络模型。将模型放在ORL人脸数据库中,通过调整网络的参数进行对比实验。实验结果表明,与BP神经网络和自组织RBF神经网络相比,弹性RBF神经网络具有更高的学习效率和识别效果。
To improved the performance of RBF neural network in face recognition, a new face recognition method was designed by applying elastic RBF(E-RBF) neural network. This method is based on neuron activity to divide the neurons, adjust the connections between neurons according to the neural repair criterion, and then change the implicit layer neuron structure and the connection weight between the hidden layer and the output neuron, and obtain an elastic RBF neural network model. We can put the model in the ORL face database, and adjust the parameters of the network to compare experiments. Finally, the experiment results show that the E-RBF neural network has higher learning efficiency and recognition effect than the BP neural network and the normal self-organizing RBF neural network algorithm.
To improved the performance of RBF neural network in face recognition, a new face recognition method was designed by applying elastic RBF(E-RBF) neural network. This method is based on neuron activity to divide the neurons, adjust the connections between neurons according to the neural repair criterion, and then change the implicit layer neuron structure and the connection weight between the hidden layer and the output neuron, and obtain an elastic RBF neural network model. We can put the model in the ORL face database, and adjust the parameters of the network to compare experiments. Finally, the experiment results show that the E-RBF neural network has higher learning efficiency and recognition effect than the BP neural network and the normal self-organizing RBF neural network algorithm.
引文
[1]李武军,王崇俊,张炜,等.人脸识别研究综述[J].模式识别与人工智能,2006,(2):58-66.
[2]张敏,徐启华.基于改进BP的神经网络模型参考自适应控制[J].软件,2015,36(7):118-123
[3]石黎,孙志梅.教学质量评价的BP神经网络模型[J].软件,2015,36(3):19-21
[4]张昭昭,乔俊飞,余文.多层自适应模块化神经网络结构设计[J].计算机学报,2017,40(12):2827-2838.
[5]乔俊飞,韩红桂.前馈神经网络分析与设计[M].北京:科技出版社,2013.
[6]GILBERT P M,HAVENSTRITE K L,MAGNUSSON K E G,et al.Substrate elasticity regulates skeletal muscle stem cell self-renewal in culture[J].Science,2012,329(5995):1078-1081.
[7]KRIVOV S,ULANOWICZ R E,DAHIVA A.Quantitative measures of organization for multiagent systems[J].Bio-systems,2003,69(1):39-54.
[8]KRIVOV S,ULANOWICZ R E,DAHIVA A.Quantitative measures of organization for multiagent systems[J].Bio-systems,2003,69(1):39-54.
[9]安大海,蒋砚军.基于BP神经网络的人脸识别系统[J].软件,2015,36(12):76-79.
[2]张敏,徐启华.基于改进BP的神经网络模型参考自适应控制[J].软件,2015,36(7):118-123
[3]石黎,孙志梅.教学质量评价的BP神经网络模型[J].软件,2015,36(3):19-21
[4]张昭昭,乔俊飞,余文.多层自适应模块化神经网络结构设计[J].计算机学报,2017,40(12):2827-2838.
[5]乔俊飞,韩红桂.前馈神经网络分析与设计[M].北京:科技出版社,2013.
[6]GILBERT P M,HAVENSTRITE K L,MAGNUSSON K E G,et al.Substrate elasticity regulates skeletal muscle stem cell self-renewal in culture[J].Science,2012,329(5995):1078-1081.
[7]KRIVOV S,ULANOWICZ R E,DAHIVA A.Quantitative measures of organization for multiagent systems[J].Bio-systems,2003,69(1):39-54.
[8]KRIVOV S,ULANOWICZ R E,DAHIVA A.Quantitative measures of organization for multiagent systems[J].Bio-systems,2003,69(1):39-54.
[9]安大海,蒋砚军.基于BP神经网络的人脸识别系统[J].软件,2015,36(12):76-79.