Is my CNN Correct

Is my CNN Correct ( It works in my data but I need to confirm it with expert review)

% CNN Definition
layers = [
    sequenceInputLayer(1)
    convolution1dLayer(5,16,'Padding','same')
    batchNormalizationLayer
    reluLayer
    convolution1dLayer(3,32,'Padding','same')
    batchNormalizationLayer
    reluLayer
    globalAveragePooling1dLayer
    fullyConnectedLayer(16)
    reluLayer
    fullyConnectedLayer(2)
    softmaxLayer
    classificationLayer
    ];

options = trainingOptions('adam', ...
    'InitialLearnRate',0.001, ...
    'MaxEpochs',30, ...
    'MiniBatchSize',128, ...
    'Shuffle','every-epoch', ...
    'Verbose',false, ...
    'ExecutionEnvironment','auto', ...
    'Plots','none');

net = trainNetwork(XTrain, categorical(YTrain), layers, options);

Answer 1

[Chuguang Pan]

You can use analyzeNetwork fucntion to obtain more details of your neural network

layers = [
            sequenceInputLayer(1)
            convolution1dLayer(5,16,'Padding','same')
            batchNormalizationLayer
            reluLayer
            convolution1dLayer(3,32,'Padding','same')
            batchNormalizationLayer
            reluLayer
            globalAveragePooling1dLayer
            fullyConnectedLayer(16)
            reluLayer
            fullyConnectedLayer(2)
            softmaxLayer
        ]; % classificationLayer is not recommended
cnnNet = dlnetwork(layers);

input = dlarray(rand(1,20,1024),"CBT");
% analyzeNetwork(cnnNet,input) % analyzeNetwork is not supported in online

Answer 2

[Walter Roberson]

There are definitely more than 10 possible layers at any point.

Your system has 12 layers.

Therefore there are over 10^12 possible combinations for 12 layers.

We have no possible way of knowing whether your one network is "the" correct network out of 10^12 possibilities.

The most we could talk about is whether it is a plausible arrangement of layers for some purpose. There is absolutely no way for us to know that it is the one true network... since you did not even describe your needs.