摘要
高阶感知器是神经元状态变量的非线性化 ,它是一阶感知器的非线性推广 ,除了神经元状态变量的非线性化推广外 ,还对权向量函数的非线性推广而得到的感知器 ,文中定义为具有非线性权向量函数的感知器 ,由于感知器的权重及作用函数都是非线性函数 ,当感知器接近最优点时 ,其连接权调节幅度很小 ,采用对非线性权函数及非线性作用函数分别进行Taylor展开 ,并取其一阶式近似逼近原函数 ,从而使其非线性权函数及非线性作用函数都转化为线性函数 ,简化了感知器学习过程的计算量 ,加快了感知器的学习过程。最后 。
High order perceptron in the nonlinear ization of neuron state variable;therefore it is the first order perceptron of nonlinear popularization.Besides the nonlinear popularization of neuron state variable,we also have the popularized perceptron of weight vector function.Based on these,the concepts of nonlinear weight function perceptron are firstly proposed.Because of the perceptron of weight vector function is nonlinear function,when perceptron weights that varies in small range at that time are near the optimal values.So the nonlinear processing elements are expanded by Taylor series and expressed approximately by a first order Taylor series.A new learning procedure which is called faster linearization learning algorithm is presented.The algorithm simplifies the perceptron training process and speeds up the perceptron training.
出处
《计算机应用与软件》
CSCD
北大核心
2004年第1期72-74,共3页
Computer Applications and Software
基金
广西自然科学基金资助项目 (桂科基 0 1 4 1 0 34)
