How does the “ trainNetwork” function define input training data?

2 次查看（过去 30 天）

cui,xingxing 2019-11-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/489114-how-does-the-trainnetwork-function-define-input-training-data

No matter how I change the input data format of image3dInputLayer, custom datastore, table, etc., can't be effectively trained?

load a.mat
layers = [
    image3dInputLayer([64 64 64 4],"Name","image3dinput")
    convolution3dLayer([11 11 7],96,"Name","conv3d","BiasLearnRateFactor",2,"Padding",[1 1 1;1 1 1],"Stride",[4 4 7])
    reluLayer("Name","relu1")
    crossChannelNormalizationLayer(5,"Name","norm1","K",1)
    averagePooling3dLayer([3 3 1],"Name","avgpool3d","Stride",[2 2 1])
    fullyConnectedLayer(8,"Name","fc")
    softmaxLayer("Name","softmax")
    classificationLayer("Name","classoutput")];
options = trainingOptions('sgdm', ...
    'MiniBatchSize',2, ...
    'MaxEpochs',2, ...
    'InitialLearnRate',3e-4, ...
    'Shuffle','every-epoch', ...
    'Plots','training-progress');
[net, tr] = trainNetwork(a,layers,options);