Adversarial AutoEncoder

• Code for the paper

Requirements

• Chainer 2+

• Python 2 or 3

Incorporating Label Information in the Adversarial Regularization

run semi-supervised/regularize_z/train.py

We trained with a prior (a mixture of 10 2-D Gaussians or Swissroll distribution) on 10K labeled MNIST examples and 40K unlabeled MNIST examples.

Supervised Adversarial Autoencoders

run supervised/learn_style/train.py

Semi-Supervised Adversarial Autoencoders

run semi-supervised/classification/train.py

| data | # | |:--:|:--:| | labeled | 100 | | unlabeled | 49900 | | validation | 10000 |

Validation accuracy at each epoch

Analogies

Unsupervised clustering

run unsupervised/clustering/train.py

16 clusters

32 clusters

Dimensionality reduction

run unsupervised/dim_reduction/train.py

run semi-supervised/dim_reduction/train.py