神经网络简单应用

clear all

clc

x=[1 2 3 4 5]

t=[1 3 5 7 9]

net=feedforwardnet(10);

net=train(net,x,t)

gensim(net)

 

net =

 

    Neural Network

 

              name: 'Feed-Forward Neural Network'%建立BP网络

          userdata: (your custom info)

 

    dimensions:%属性

 

         numInputs: 1%

         numLayers: 2

        numOutputs: 1

    numInputDelays: 0

    numLayerDelays: 0

 numFeedbackDelays: 0

 numWeightElements: 31

        sampleTime: 1

 

    connections:

 

       biasConnect: [1; 1]

      inputConnect: [1; 0]

      layerConnect: [0 0; 1 0]

     outputConnect: [0 1]

 

    subobjects:

 

             input: Equivalent to inputs{1}

            output: Equivalent to outputs{2}

 

            inputs: {1x1 cell array of 1 input}

            layers: {2x1 cell array of 2 layers}

           outputs: {1x2 cell array of 1 output}

            biases: {2x1 cell array of 2 biases}

      inputWeights: {2x1 cell array of 1 weight}

      layerWeights: {2x2 cell array of 1 weight}

 

    functions:

 

          adaptFcn: 'adaptwb'

        adaptParam: (none)

          derivFcn: 'defaultderiv'

         divideFcn: 'dividerand'

       divideParam: .trainRatio, .valRatio, .testRatio

        divideMode: 'sample'

           initFcn: 'initlay'

        performFcn: 'mse'

      performParam: .regularization, .normalization

          plotFcns: {'plotperform', plottrainstate, ploterrhist,

                    plotregression}

        plotParams: {1x4 cell array of 4 params}

          trainFcn: 'trainlm'

        trainParam: .showWindow, .showCommandLine, .show, .epochs,

                    .time, .goal, .min_grad, .max_fail, .mu, .mu_dec,

                    .mu_inc, .mu_max

 

    weight and bias values:

 

                IW: {2x1 cell} containing 1 input weight matrix

                LW: {2x2 cell} containing 1 layer weight matrix

                 b: {2x1 cell} containing 2 bias vectors

 

    methods:

 

             adapt: Learn while in continuous use

         configure: Configure inputs & outputs

            gensim: Generate Simulink model

              init: Initialize weights & biases

           perform: Calculate performance

               sim: Evaluate network outputs given inputs

             train: Train network with examples

              view: View diagram

       unconfigure: Unconfigure inputs & outputs