首页>
外国专利>
LEARNING METHOD OF CHOAS CIRCULAR NEURAL NET
LEARNING METHOD OF CHOAS CIRCULAR NEURAL NET
展开▼
机译:混沌圆神经网络的学习方法
展开▼
页面导航
摘要
著录项
相似文献
摘要
The present invention discloses a learning method in a cyclic neural network using a chaotic neural network. The method comprises the steps of: initializing the initial condition x i (0) and the connection weight to a random value; A second step of performing a network according to Equation (1) during a period T according to the initial condition x i (0) and an external input a i (t);;(here, , w ij is the weight of the connection from the j th neuron to the i th neuron, f i is the output function of the i th neuron, y i (t) and x i (t) are the time-varying internal states and outputs of the i th neuron, a i (t) is the external input of the i-th neuron.) L i (T) = 0, S x i (0) = 0 Lagrange multiplier L i by the given teacher signal Q i (t) a third step of obtaining (t) inversely from the time T according to Equation (2) below, wherein x i (t) uses the value calculated in the second step;;Where δE is the change in total error and δw ij is the small change in link weight.;A fourth step of performing a process from the second step to the third step on all inputs and calculating a sum of total errors; And a fifth step of repeating learning from the second step after correcting the weight by the following equation (3), which is a weight correction equation if the error is less than or equal to a predetermined limit error in the fourth step;;Consists of
展开▼
机译:本发明公开了一种使用混沌神经网络的循环神经网络中的学习方法。该方法包括以下步骤:将初始条件x i Sub>(0)初始化,并将连接权重初始化为随机值;第二步,根据初始条件x i Sub>(0)和外部输入a i Sub>(t)在时间段T内根据等式(1)执行网络;;(这里,w ij Sub>是从第j个神经元到第i个神经元的连接权重,f i Sub>是第i个神经元的输出函数,y i Sub>(t)和x i Sub>(t)是第i个神经元 i Sub>的时变内部状态和输出(t)是第i个神经元的外部输入。)L i Sub>(T)= 0,S x i Sub>(0)= 0拉格朗日乘数L i Sub>通过给定的教师信号Q i Sub>(t)从下面的等式(2)开始,从时间T反过来获得(t)的第三步,其中x i Sub>(t)使用第二步中计算出的值;δE是总误差的变化,而δw ij Sub>是链路权重的小变化。所有输入和计算从第二步到第三步的过程ng总误差之和;第五步,在通过以下方程式(3)校正权重之后,从第二步重复学习,如果第四步中的误差小于或等于预定极限误差,则为权重校正方程;
展开▼