How do I draw the MSE curve for an LSTM classification code in Matlab?

I have a classification code for LSTM in Matlab and I want to draw the curve of MSE for testing data and there is an error appear when I implement the code of MSE=mean(y_test-y_pred)^2.

How can I draw this curve please?

Answer 1

[Manish]

Hi Ruaa,

I understand that you would like to plot the Mean Squared Error (MSE) curve for the test data. I've used a random sequence dataset and implemented an LSTM model for demonstration purposes.

Refer to the sample code below for better understanding.

numObservations = 100;
sequenceLength = 10;   
numFeatures = 5;       
numClasses = 3;        
%random sequences for training
XTrain = arrayfun(@(x) randn(numFeatures, sequenceLength), 1:numObservations, 'UniformOutput', false);
YTrain = categorical(randi(numClasses, numObservations, 1));

%random sequences for testing
XTest = arrayfun(@(x) randn(numFeatures, sequenceLength), 1:numObservations, 'UniformOutput', false);
YTest = categorical(randi(numClasses, numObservations, 1));

% LSTM network
layers = [
    sequenceInputLayer(numFeatures)
    lstmLayer(50, 'OutputMode', 'last')
    fullyConnectedLayer(numClasses)
    softmaxLayer
    classificationLayer];

% Training options
maxEpochs = 5;
options = trainingOptions('adam', ...
    'MaxEpochs', maxEpochs, ...
    'MiniBatchSize', 16, ...
    'SequenceLength', 'longest', ...
    'Shuffle', 'every-epoch', ...
    'Verbose', 0);

% Initialize MSE storage
msePerEpoch = zeros(maxEpochs, 1);

% Train and calculate MSE per epoch
for epoch = 1:maxEpochs
    net = trainNetwork(XTrain, YTrain, layers, options);
    YPred = classify(net, XTest, 'SequenceLength', 'longest');
    % Convert categorical predictions and true labels to numeric
    y_pred = double(YPred);
    y_test = double(YTest);
    % Calculate MSE for this epoch
    msePerEpoch(epoch) = mean((y_test - y_pred).^2);
end

% Plot MSE against epochs
figure;
plot(1:maxEpochs, msePerEpoch, '-o');
title('MSE Over Epochs');
xlabel('Epoch');
ylabel('MSE');
grid on;

●

Refer the below documentation link for better understanding:

• https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.lstmlayer.html

Hope it helps!