swang373 / SIG_GIS

Sliced Iterative Generator (SIG) & Gaussianizing Iterative Slicing (GIS)

Sliced Iterative Generator (SIG) & Gaussianizing Iterative Slicing (GIS)

This repository is the official implementation of Sliced Iterative Generator.

Requirements

To install requirements:

pip install -r requirements.txt

Training

To train SIG and GIS, run the following commands:

python SIG.py --dataset DATASET --seed SEED --save SAVING_ADDRESS   
python GIS.py --dataset DATASET --seed SEED --save SAVING_ADDRESS

Evaluation

All the evaluation codes of SIG are in notebook/SIG.ipynb. The FID score of SIG samples can also be evaluated on the fly with the "--evaluateFID" argument.

Results

SIG achieves the following performance on :

FID score

MNIST	Fashion	CIFAR10	CelebA
4.5	13.7	66.5	37.3

OoD detection (AUROC on models trained on FashionMNIST)

MNIST	OMNIGLOT	FMNIST-hflip	FMNIST-vflip
0.990	0.993	0.631	0.821

The codes to reproduce these results are in SIG.ipynb