Predictors must be a numeric array, a datastore, or a table.

clear all; close all; clc;

load mfcc_S6_8_01.mat;

load mfcc_S6_8_02.mat;

load mfcc_S6_9_01.mat;

load mfcc_S6_9_02.mat;

[s1 s2]=size(sg_S6_8_01);

%YTrain_data=[sg_S6_8_01 sg_S6_8_02 sg_S6_9_01 sg_S6_9_02];

cmb_8_01=[];

for ii=1:s2

cmb_sg_8=sg_S6_8_01{ii};

cmb_8_01=[cmb_8_01; cmb_sg_8];

end

[s1 s2]=size(sg_S6_8_02)

cmb_8_02=[];

for ii=1:s2

cmb_sg_8_2=sg_S6_8_02{ii};

cmb_8_02=[cmb_8_02; cmb_sg_8_2];

end

[s1 s2]=size(sg_S6_9_01)

cmb_9_01=[];

for ii=1:s2

cmb_sg_9_01=sg_S6_9_01{ii};

cmb_9_01=[cmb_9_01; cmb_sg_9_01];

end

[s1 s2]=size(sg_S6_9_02)

cmb_9_02=[];

for ii=1:s2

cmb_sg_9_02=sg_S6_9_02{ii};

cmb_9_02=[cmb_9_02; cmb_sg_9_02];

end

comb_train_data={cmb_8_01;cmb_8_02;cmb_9_01;cmb_9_02;};

index2=1;

files={sg_S6_8_01; sg_S6_8_02;sg_S6_9_01;sg_S6_9_02};

for ii=1:length(files)

chk_files=files{ii};

for jj=1:length(chk_files)

comb_data=chk_files{jj};

all_frames{:,:,index2}=comb_data;

index2=index2+1;

end

end

YTrain_lbl=categorical([8*ones(1,50) 8*ones(1,51) 9*ones(1,47) 9*ones(1,47)]);

%YTrain_lbl=categorical(YTrain1_lbl);

for kk=1:195

if (kk<=1 && kk>=101)

label1{kk}=8;

else

label1{kk}=9;

end

end

YTrain=all_frames;

classWeights = 1./countcats(YTrain_lbl);

classWeights = classWeights'/mean(classWeights);

numClasses = numel(categories(YTrain_lbl));

numHops=129;numBands=7;

timePoolSize = ceil(numHops/8);

dropoutProb = 0.2;

numF = 12;

layers = [

imageInputLayer([numHops numBands])

convolution2dLayer(3,numF,'Padding','same')

batchNormalizationLayer

reluLayer

maxPooling2dLayer(3,'Stride',2,'Padding','same')

convolution2dLayer(3,2*numF,'Padding','same')

batchNormalizationLayer

reluLayer

maxPooling2dLayer(3,'Stride',2,'Padding','same')

convolution2dLayer(3,4*numF,'Padding','same')

batchNormalizationLayer

reluLayer

maxPooling2dLayer(3,'Stride',2,'Padding','same')

convolution2dLayer(3,4*numF,'Padding','same')

batchNormalizationLayer

reluLayer

convolution2dLayer(3,4*numF,'Padding','same')

batchNormalizationLayer

reluLayer

maxPooling2dLayer([timePoolSize,1])

dropoutLayer(dropoutProb)

fullyConnectedLayer(numClasses)

softmaxLayer

weightedClassificationLayer(classWeights)];

miniBatchSize = 128;

%validationFrequency = floor(numel(YTrain)/miniBatchSize);

options = trainingOptions('adam', ...

'InitialLearnRate',3e-4, ...

'MaxEpochs',25, ...

'MiniBatchSize',miniBatchSize, ...

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

'Plots','training-progress', ...

'Verbose',false, ...

'LearnRateSchedule','piecewise', ...

'LearnRateDropFactor',0.1, ...

'LearnRateDropPeriod',20)

trainedNet = trainNetwork(all_frames,label1',layers,options);

ERROR:

Error using trainNetwork (line 184)

Invalid training data. Predictors must be a numeric array, a datastore, or a table. For networks with

sequence input, predictors can also be a cell array of sequences.

Error in JAN31_2023_EX (line 113)

trainedNet = trainNetwork(all_frames,label1',layers,options);

Sir/ Madam

I have written code for digit recognition. Input is MFCC matrix only for digit 8 & 9 in terms of frames. In this code, digit 8 frames are 101 and digit 9 frames are 94 frames. i have declared label variable according to that. Please guide to resolve above error

Answer 1

[Walter Roberson]

your responses in ylabel1 is a cell array of scalar numbers. cell array with numeric content is supported only for sequence to sequence regression, but your input data is not one of the types that sequences can be coded as. You should probably be using a pure numeric array.

https://www.mathworks.com/help/deeplearning/ref/trainnetwork.html#mw_d0b3a2e4-09a0-42f9-a273-2bb25956fe66

if (kk<=1 && kk>=101)

There is no number that is simultaneously less than or equal to 1 but also greater than or equal to 101. You need to rethink that test.

[Shilpa Sonawane]

Sir,

In the code, ylabel1 is cell array with numbers. Could you please guide about decleration of cell array with numeric content.

Thank you sir.