Invalid training data. X and Y must have the same number of observations.

Good morning,

I'm trying to train a neural network but "Invalid training data. X and Y must have the same number of observations." error came out. I am not able to understand where is the problem.

XTrain is a 24x4x1x1321 double and loadYTrain is 1321x1 double (I have also tryed to train with its transpose: 1x1321).

Can you please help me out, it's very important.

Thank you very much to all in advance.

Here is my code:

layers=[...

sequenceInputLayer([24 4 1],'Name','input')

sequenceFoldingLayer('Name','fold')

convolution2dLayer(3,32,'Padding','same','Name','2d1')

batchNormalizationLayer('Name','bn1')

reluLayer('Name','relu1')

maxPooling2dLayer(2,'Name','pool1')

convolution2dLayer(3,64,'Padding','same','Name','2d2')

reluLayer('Name','relu2')

batchNormalizationLayer('Name','bn2')

maxPooling2dLayer(2,'Name','pool2')

convolution2dLayer(3,128,'Padding','same','Name','2d3')

reluLayer('Name','relu3')

batchNormalizationLayer('Name','bn3')

maxPooling2dLayer(2,'Name','pool3')

convolution2dLayer(3,256,'Padding','same','Name','2d4')

reluLayer('Name','relu4')

batchNormalizationLayer('Name','bn4')

sequenceUnfoldingLayer('Name','unfold')

flattenLayer('Name','flat')

lstmLayer(4,'Name','lstm1','OutputMode','sequence')

dropoutLayer(0.5,'Name','drop1')

lstmLayer(8,'Name','lstm2','OutputMode','sequence')

dropoutLayer(0.5,'Name','drop2')

lstmLayer(16,'Name','lstm3','OutputMode','sequence')

dropoutLayer(0.5,'Name','drop3')

lstmLayer(32,'Name','lstm4','OutputMode','sequence')

dropoutLayer(0.5,'Name','drop4')

fullyConnectedLayer(1,'Name','full')

reluLayer('Name','relu6')

regressionLayer('Name','reg')];

maxEpochs = 30;

miniBatchSize = 128;

options = trainingOptions('sgdm', ...

'MiniBatchSize',miniBatchSize, ...

'MaxEpochs',maxEpochs, ...

'InitialLearnRate',1e-3, ...

'LearnRateSchedule','piecewise', ...

'LearnRateDropFactor',0.1, ...

'LearnRateDropPeriod',20, ...

'Shuffle','every-epoch', ...

'Verbose',false);

lgraph = layerGraph(layers);

lgraph = connectLayers(lgraph,'fold/miniBatchSize','unfold/miniBatchSize');

net = trainNetwork(XTrain,loadYTrain,lgraph,options);

Answer 1

[Srivardhan Gadila]

In order to model the LSTM regression networks with 2-D data, the input should be a Nx1 cell, N being the number of observations. Each cell entry should then comprise a HxWxCxS array, where H = height, W=width, C=channels and S= sequence length. The responses, Y, can be a NxR matrix, where N = observations, R = number of responses (or output of the network) for sequence-to-one problems or a Nx1 cell array of RxS responses for sequence-to-sequence problems.

Refer to Input Arguments of trainNetwork, then check sequences input argument explanation on what should be the format and shape of XTrain & Y input argument explanation on what should be the format and shape of the YTrain.

The following is a similar Question: CNN and LSTM error with input size

[Francesco Laezza]

Thank you very much, I managed to solve my problem. Now it seems to train the network but does not come out the classical window that show the training ongoing. What could be the problem?

[Srivardhan Gadila]

Refer to Plots and Display and add the name value pair arguments 'Plots','training-progress' to the trainingOptions

[Francesco Laezza]

Thank you very much, you were helpful!