Skip to content

Using GANs to Relax the Bottleneck of Curated Labeled Data, ArxXiv paper: https://arxiv.org/pdf/1803.05137.pdf, CVPR 2018.

Notifications You must be signed in to change notification settings

ArghyaPal/Adversarial_Data_Programming

Repository files navigation

Getting Started

Please see the adversarial data programming website page to make yourself comfortable with the adversarial data programming concept. We will train a basic generative adversarial network (GAN) on MNIST digit dataset - using labeling functions. We will do the following steps:

  1. Model a basic generative adversarial network with linear layers
  2. We will formalise our Labeling Functions Block using a labeling function based on Kmeans clustering
  3. We will train the model end-to-end on MNIST dataset

Installation

You need to install:

  1. Pytorch > 1.2+
  2. Python 3.6+
  3. matplotlib
  4. torchvision
  5. install kmeans of pytorch using the command pip install kmeans-pytorch
  6. pylab

Results

The generated images are stored in samples folder. However, you can see the labels after every iteration

Acknowledgement

The GAN code is based on . And we thank kmeans-pytorch for providing the kmeans unsupervised clustering code.

How to write labeling functions for real image datasets

You can see the tutorial to write labeling functions for real dataset.

Citation

If you find the code useful, please cite the paper:
@InProceedings{Pal_2018_CVPR, author = {Pal, Arghya and Balasubramanian, Vineeth N.}, title = {Adversarial Data Programming: Using GANs to Relax the Bottleneck of Curated Labeled Data}, booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, month = {June}, year = {2018} }

About

Using GANs to Relax the Bottleneck of Curated Labeled Data, ArxXiv paper: https://arxiv.org/pdf/1803.05137.pdf, CVPR 2018.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published