Spectral Normalization for Generative Adversarial Networks
Machine Learning
2018-02-19 v1 Computer Vision and Pattern Recognition
Machine Learning
Abstract
One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs (SN-GANs) is capable of generating images of better or equal quality relative to the previous training stabilization techniques.
Cite
@article{arxiv.1802.05957,
title = {Spectral Normalization for Generative Adversarial Networks},
author = {Takeru Miyato and Toshiki Kataoka and Masanori Koyama and Yuichi Yoshida},
journal= {arXiv preprint arXiv:1802.05957},
year = {2018}
}
Comments
Published as a conference paper at ICLR 2018