philipperemy / EGAIN-pytorch

Generative Adversarial Imputation Networks (GAIN)

Pytorch implementation of EGAIN, an extension of GAIN.

Refer to this link for the official implementation in Tensorflow V1.

This directory contains implementations of GAIN framework for imputation using two UCI datasets.

• UCI Letter (https://archive.ics.uci.edu/ml/datasets/Letter+Recognition)
• UCI Spam (https://archive.ics.uci.edu/ml/datasets/Spambase)

Command inputs

• data_name: letter or spam
• miss_rate: probability of missing components
• batch_size: batch size
• hint_rate: hint rate
• alpha: hyperparameter
• iterations: iterations

Example command

$ python3 main_letter_spam.py --data_name spam --miss_rate 0.2 --batch_size 128 --hint_rate 0.9 --alpha 100 --iterations 10000