ABCAS: Adaptive Bound Control of spectral norm as Automatic Stabilizer

Shota Hirose, Shiori Maki, Naoki Wada, Jiro Katto, Heming Sun

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Spectral Normalization is one of the best methods for stabilizing the training of Generative Adversarial Network. Spectral Normalization limits the gradient of discriminator between the distribution between real data and fake data. However, even with this normalization, GAN's training sometimes fails. In this paper, we reveal that more severe restriction is sometimes needed depending on the training dataset, then we propose a novel stabilizer which offers an adaptive normalization method, called ABCAS. Our method decides discriminator's Lipschitz constant adaptively, by checking the distance of distributions of real and fake data. Our method improves the stability of the training of Generative Adversarial Network and achieved better Fréchet Inception Distance score of generated images. We also investigated suitable spectral norm for three datasets. We show the result as an ablation study.

Original languageEnglish
Title of host publication2023 IEEE International Conference on Consumer Electronics, ICCE 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665491303
DOIs
Publication statusPublished - 2023
Event2023 IEEE International Conference on Consumer Electronics, ICCE 2023 - Las Vegas, United States
Duration: 2023 Jan 62023 Jan 8

Publication series

NameDigest of Technical Papers - IEEE International Conference on Consumer Electronics
Volume2023-January
ISSN (Print)0747-668X

Conference

Conference2023 IEEE International Conference on Consumer Electronics, ICCE 2023
Country/TerritoryUnited States
CityLas Vegas
Period23/1/623/1/8

Keywords

  • Adaptive Scheduling
  • Generative Adversarial Network
  • Lipschitz constant
  • Spectral Normalization

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'ABCAS: Adaptive Bound Control of spectral norm as Automatic Stabilizer'. Together they form a unique fingerprint.

Cite this