STFRFT

Matlab Guide [Korean Vision]

说明：  瑞利信道瑞利信道瑞利信道瑞利信道瑞利信道瑞利信道瑞利信道(Rayleigh channel)

Speech oscilloscope using matlab

Elman神经网络：Elman网络是 J. L. Elman于1990年首先针对语音处理问题而提出来的, 它是一种典型的局部回归网络( global feed for ward l ocal recurrent)。Elman网络可以看作是一个具有局部记忆单元和局部反馈连接的前向神经网络。Elman网络具有与多层前向网络相似的多层结构。它的主要结构是前馈连接, 包括输入层、 隐含层、 输出层, 其连接权可以进行学习修正 反馈连接由一组“结构 ” 单元构成,用来记忆前一时刻的输出值, 其连接权值是固定的。在这种网络中, 除了普通的隐含层外, 还有一个特别的隐含层,称为关联层 (或联系单元层 ) 该层从隐含层接收反馈信号, 每一个隐含层节点都有一个与之对应的关联层节点连接。关联层的作用是通过联接记忆将上一个时刻的隐层状态连同当前时刻的网络输入一起作为隐层的输入, 相当于状态反馈。隐层的传递函数仍为某种非线性函数, 一般为 Sigmoid函数, 输出层为线性函数, 关联层也为线性函数。(Elman neural network: Elman JL Elman network is first proposed for speech processing problems in 1990. It is a typical local regression network (global feed for ward l ocal recurrent). Elman network can be seen as having a forward neural network of local memory unit and local feedback connections. Elman network has a multi-layer multilayer structure similar to the network. Its main structure is a feedforward connection, comprising an input layer, a hidden layer and output layer, which can be connected right learning correction feedback connected by a set of " structural" units, the output value of the memory for the previous time, which connection weights are fixed. In such a network, in addition to ordinary hidden layer, there is a special hidden layer, called an associative layer (or contact layer unit) this layer receives feedback signals the hidden layer, each hidden layer node has an associated layer node corresponding connection. The role of the associated layer is fo)

用遗传算法优化bp网络，实现的方式很新颖也很成功，实现了网络的不陷入局部收敛(Using genetic algorithm to optimize the bp network, is very innovative and successful implementation, realize the network do not fall into local convergence)

Error calculation set. For statictic purposes

matlab程序：用最小二乘法拟合椭球曲面(matlab program: using least-squares fitting ellipsoid surface)

径向基函数神经网络，资源分配神经网络，可以设计出结构最精简的径向基函数神经网络(Radial basis function neural network, resource allocation, neural network, we can design the most streamlined structure of radial basis function neural network )

Berger algorythm heart rate variability by me(Eegii33)

解决M/M/1排队论问题，服务时间服从指数分布，队列为泊松流。系统输出为每个个体的到达时间，离开时间，等待时间，逗留时间(M/M/1 queuing theory to solve the problem, the service time is exponentially distributed, the queue is Poisson. System output for each individual arrival time, departure time, waiting time, length of stay)