execution instuctions: $ python3 autoencoder.py -d <dataset_path>
The program can't be executed without the path of the dataset as an argument.
The program trains an autoencoder to encode and reconstruct images.
The dataset used for the project is the MNIST dataset of hand written digits.
Best results with hyperparameters:
layer_num = 5
filter_size = 64
filter_num = 3
epochs = 80
batch_size = 256
