Adversarial autoencoder (basic/semi-supervised/supervised)
First,
$ python create_datasets.py
Then, you get data/MNIST, data/subMNIST (atomatically downloded in data/ directory), which are MNIST image datasets. you also get train_labeled.p, train_unlabeled.p, validation.p, which are list of tr_l, tr_u, tt image.
Second,
$ python aae_baisc.py
or
$ python aae_supervised.py
You can get category conditional images like below,
$ python aae_semi_supervised.py