Are the training and testing images in the example train stacked autoencoder the same?

https://www.mathworks.com/help/nnet/examples/train-stacked-autoencoders-for-image-classification.html

in the above example are they considering the same training set data as the testing data with jumbled sequence?or are they using 2 different datasets? can anybody please answer this

[CHHAVI]

How can i use cross validation in this example instead of dividing data in training and testing?

Answer 1

[Vishal Bhutani]

Based on my understanding you want to know about the dataset for training and testing in stacked autoencoder. The datasets used for training and testing are different. They both are synthetic dataset, synthetic images have been generated by applying random affine transformations to digit images created using different fonts. And as you may aware, in Neural Networks both training and testing data is different. And as mentioned in the code while loading training and testing data separate dataset is loaded.

% Load the training data into memory
[xTrainImages,tTrain] = digitTrainCellArrayData;

% Load the test images
[xTestImages,tTest] = digitTestCellArrayData;

Hope it might help you.

[debojit sharma]

Suppose I have 4 different classes of images organized in separate folders for each class. How can I prepare these dataset to fed to a stacked autoencoder? Like in this example how 'digitTrainCellArrayData' dataset is prepared so as to make it ready to fed in a autoencoder. Please provide a sample code for this