Home

Awesome

UDA : Unsupervised Data Augmentation

Unofficial PyTorch Implementation of Unsupervised Data Augmentation.

Most of codes are from Fast AutoAugment.

Introduction

todo.

Run

$ python train.py -c confs/wresnet28x2.yaml --unsupervised

Experiments

Cifar10 (Reduced, 4k dataset)

Reproduce Paper's Result

WResNet 28x2PaperOur Converged(Top1 Err)Our Best(Top1 Err)
Supervised20.2621.30
AutoAugment14.1*15.413.4
UDA5.276.586.27

SVHN

todo.

ImageNet

todo.

References