Feature Extraction using deep autoencoder

I have filtered my ecg signal of 108000*1 length and then divided into blocks using window size of 64 samples each. Now i need to extract feature from each window using deep autoencoder in MATLAB. any help or idea how can i perform this? Thanks in advance.

Answer 1

[BERGHOUT Tarek]

1) you must create a data set of this windows , dataset =[window1;window2; window3 ...................].

2) train these dataset with an AES.

3) the hidden layer will be your new extructed dataset;

[Shankar Parmar]

Sir,

How can I extract this Hidden Layer in MATLAB using

trainAutoencoder command.

[arahiche]

To access the extracted features you need to use encode function.

here is an example;

hiddenSize = 100; % for example
AE_model = trainAutoencoder(Input_data,hiddenSize);
% you can view you model using this function
view(AE_model)
% To access the latent code generated 
features = encode(AE_model,Input_data);