Semi-Supervised Learning with GANs: Revisiting Manifold Regularization

Code

• github.com/bruno-31/GAN-manifold-regularization
  OfficialIn papertf★ 0
• github.com/UCI-ML-course-team/GAN-manifold-regularization-PyTorch
  pytorch★ 0

Abstract

GANS are powerful generative models that are able to model the manifold of natural images. We leverage this property to perform manifold regularization by approximating the Laplacian norm using a Monte Carlo approximation that is easily computed with the GAN. When incorporated into the feature-matching GAN of Improved GAN, we achieve state-of-the-art results for GAN-based semi-supervised learning on the CIFAR-10 dataset, with a method that is significantly easier to implement than competing methods.

Reproductions