Ver_ENG
this python script is autoencoder for mnist datasets and
simple autoencoder that didn't use CNN(Convolution Neural Network.)
but it is pretty good worked.(only simple data.)
more information for this mnist datasets see this site.
tensorflow site : https://www.tensorflow.org/tutorials/keras/basic_classification
python training [flatted | normal]
arguments
[flatted] : flat original data(2D [28, 28]) to 1D data[28*28].
[normal] use origin data shape[28, 28]
notice! I don't used code for weight saving(a.k.a tf.saver) if you want to save weights. insert code for saving data!
low performance(can't recognize feature of clothes in detail. Only the outline of clothes has been learned.)
using [normal] argument
better performance than using flatted arguments
autoencoder.py is an unused file. please using the other files. actually mnist_autoencoder.py is made of autoencoder.py. (it is a file that has been structured)