how to define sequence input layer and fully connected layer for CNN for multivariate data in matab

9 次查看（过去 30 天）

显示更早的评论

NN 2022-1-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1627250-how-to-define-sequence-input-layer-and-fully-connected-layer-for-cnn-for-multivariate-data-in-matab

评论： Pratyush Roy 2022-1-20

在 MATLAB Online 中打开

Hi,

I am trying to do CNN with two dimensional data , below is my code for layers , i am getting error .

Error using trainNetwork (line 165)

Invalid training data. Sequence responses must have the same sequence length as the corresponding

predictors.

Error in Multivariate (line 73)

net = trainNetwork(pn,tn,layers,options);

Please help me to understand the mistake:

XTrain is 1x515

YTrain is 1x515

XTest is 1x212

YTest is 1x212

below is my code :

numFeatures = 1;
numResponses = 1;
numHiddenUnits1 = 50;
FiltZise = 5;     
layers = [...
        sequenceInputLayer([numFeatures 515 1],'Name','input')
        sequenceFoldingLayer('Name','fold')
        convolution2dLayer(FiltZise,256,'Padding','same','WeightsInitializer','he','Name','conv','DilationFactor',1);
        batchNormalizationLayer('Name','bn')
        reluLayer('Name','relu')
        convolution2dLayer(FiltZise,256,'Padding','same','WeightsInitializer','he','Name','conv1','DilationFactor',2);
        reluLayer('Name','relu1')
        averagePooling2dLayer(1,'Stride',FiltZise,'Name','pool1')
        sequenceUnfoldingLayer('Name','unfold')
        flattenLayer('Name','flatten')
        fullyConnectedLayer(numResponses,'Name','fc')
        regressionLayer('Name','output')    ];
        layers = layerGraph(layers);
        layers= connectLayers(layers,'fold/miniBatchSize','unfold/miniBatchSize');
options = trainingOptions('adam','MaxEpochs',150,'MiniBatchSize',15,'GradientThreshold',1,'InitialLearnRate',0.005,'LearnRateSchedule','piecewise','LearnRateDropPeriod',125,'LearnRateDropFactor',0.2,'Verbose',0, 'Plots','training-progress');
net = trainNetwork(pn,tn,layers,options);