How to do feature extraction using wavelet scattering and then perform neural network classification?

% I want to do feature extraction for signals and then apply ANN classification on them,, can you help please?

Fs = 50; sf = waveletScattering('SignalLength', 4950, 'SamplingFrequency', Fs) ;

s = cell(120,1); parfor j = 1:120; r = featureMatrix(sf, f(j, :)) ; s{j, 1} = r; end

h = cell2mat(s) ; b = cell2table(s) ; c = table2dataset(b);

%the single signal was in one row, but after this it becomes 202rows×10columns, and I can't apply ANN like this Any ideas?

[Mustafa Mustafa]

I can also use a normal for loop

Answer 1

[Suraj Kumar]

Hi Mustafa,

To perform feature extraction using wavelet scattering on signals and then apply ANN classification, you can refer to the following steps:

1. After storing the features in cell array "s", you can convert them into a matrix "featureVectors" suitable for ANN.

% Convert cell array to feature matrix 
for i = 1:numSignals 
    featureVectors(i, :) = s{i}(:)';  
end 

2. Configure and train the neural network using ‘fitcnet’ function specifying the required parameters. % Train the neural

network net = fitcnet(featureVectors, targetsCategorical, ... 
       'LayerSizes', hiddenLayerSize, ... 
       'Standardize', true, ... 
       'Options', options); 

To learn more about ‘fitcnet’ function in MATLAB, refer the below mentioned documentation:

https://www.mathworks.com/help/stats/fitcnet.html

Hope this resolves your query!