Invalid training data. Predictors must be a N-by-1 cell array of sequences, where N is the number of sequences. All sequences must have the same feature dimension and at least

i am unable to resolve this error

clc;

close all;

clear all;

% Load dataset

load 's0486.txt'

% Open text file

fid = fopen('s0486.txt','r');

% Read text file into cell array

C = textread('s0486.txt','%s','delimiter','\n');

% Close text file

fclose(fid);

X= C(1,:);

% mydata contains a signal dataset with input features X and labels Y

%labels

Y1= cell(1,15);

Y1(1:6)={ 'normal'};

Y1(7:15)={'lvh'};

Y=categorical (Y1);

% Split dataset into training and testing sets

trainInd= [1:12];

testInd= [13:15];

XTrain = C(trainInd,:);

YTrain = Y(trainInd);

XTest = C(testInd,:);

YTest = Y(testInd);

% Define RNN architecture

inputSize = size(XTrain,1);

numHiddenUnits = 100;

outputSize = 2;

layers = [ ...

sequenceInputLayer(inputSize)

lstmLayer(numHiddenUnits,'OutputMode','last')

fullyConnectedLayer(outputSize)

softmaxLayer

classificationLayer];

% Set training options

options = trainingOptions('adam', ...

'MaxEpochs',50, ...

'MiniBatchSize',64, ...

'SequenceLength','longest', ...

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

'Verbose',false, ...

'Plots','training-progress');

% Train RNN

[net, traininfo]= trainNetwork(XTrain,YTrain,layers,options);

% Test RNN

YPred = classify(net,XTest);

accuracy = sum(YPred == YTest)/numel(YTest);

fprintf('Test accuracy: %f\n',accuracy);

Answer 1

[Nayan]

Hi,

I reproduced the exact error with a test .txt file. The code is failing at the line

net = trainNetwork(sequences,responses,layers,options)

It seems the error has occured because the "responses/YTrain(in your case)" argument of the "trainNetwork" should be a N*1 vector. However in your code it is 1*N vector.

I would refer you to the look at "Train Network for Sequence Classification" example in

Train deep learning neural network - MATLAB trainNetwork (mathworks.com) for more intution.

Also, if there are further query, it would be easier to help if you could share the ".txt" file used in the above code.