Number of observations in X and Y disagree.

Hey everyone - Cannot for the life of me get this to work. No matter what I do, I get the "Number of observations in X and Y disagree." error. I understand that X train should be a HxWxCxN 4-D matrix and that YTrain should be a Nx1 matrix. I've checked that already and this seems correct.

Any help would be appreciated.

Thanks,

Ben Payne

data = load('C:\Users\Ben Payne\Downloads\matlab\matlab\emnist-letters.mat');

testImages = reshape(data.dataset.test.images.', 28, 28, []);

figure;

for i = 1:20
    subplot(4,5,i);    
    imshow(testImages(:,:,i));
end

trainingImages = reshape(data.dataset.train.images.', 28, 28, []);

idx = randperm(size(trainingImages,3),1000);
XValidation = trainingImages(:,:,idx);
YValidation = categorical(data.dataset.train.labels(idx));

layers = [
    imageInputLayer([28 28 1])
    
    convolution2dLayer(3,8,'Padding','same')
    batchNormalizationLayer
    reluLayer
    
    maxPooling2dLayer(2,'Stride',2)
    
    convolution2dLayer(3,16,'Padding','same')
    batchNormalizationLayer
    reluLayer
    
    maxPooling2dLayer(2,'Stride',2)
    
    convolution2dLayer(3,32,'Padding','same')
    batchNormalizationLayer
    reluLayer
    
    fullyConnectedLayer(10)
    softmaxLayer
    classificationLayer
];

options = trainingOptions('sgdm', ...
    'InitialLearnRate',0.01, ...
    'MaxEpochs',8, ...
    'Shuffle','every-epoch', ...
    'ValidationData',{XValidation,YValidation}, ...
    'ValidationFrequency',30, ...
    'Verbose',false, ...
    'Plots','training-progress');

XTrain = reshape(data.dataset.train.images', 28, 28, 1, []);
YTrain = categorical(data.dataset.train.labels);

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

The dataset can be downloaded from here

http://www.itl.nist.gov/iaui/vip/cs_links/EMNIST/matlab.zip

Answer 1

[Shashank Gupta]

Hi Ben,

I am not sure if you already figured out the mistake. Neverthless it seems the YTrain has 26 classes which is conflcting with the number of neuron(10) defined at the last fully connnected layer. Ideally both should be same, I did not dig much in to the code but surely changing the neuron to 26 would solve your problem.

I hope it helps you.

Cheers.