“Cross Sample Similarity for Stable Training of GAN” by Lee and Lee

Conference:

SIGGRAPH 2021

Type:

Posters

Entry Number: 36

Title:

Cross Sample Similarity for Stable Training of GAN

Presenter(s)/Author(s):

• Jung Eun Lee
• Seungkyu Lee

Abstract:

Recently attention network finding similarity in non-local area within a 2D image has shown outstanding improvement in multi-class generation task in GAN. However it frequently shows unstable training state sometimes falling in mode collapse.We propose cross sample similarity loss to penalize similar features of fake samples that are rarely observed in reals. Proposed method shows improved FID score compared to baseline methods on CelebA, LSUN, and decreased mode collapse on Cifar10[Krizhevsky 2009].

Keyword(s):

• deep generative model
• generative adversarial nets(GAN)
• mode collapse
• autoencoder

PDF:

• Cross Sample Similarity for Stable Training of GAN

ACM Digital Library Publication:

Cross Sample Similarity for Stable Training of GAN

Overview Page:

SIGGRAPH 2021: Posters