GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- 类型:arxiv
- 标识:1706.08500
- 链接:https://arxiv.org/abs/1706.08500
- 主题:evaluation
- 主分类:llm-infra
- 形态:method
- 被引:3831
- 被引来源:OpenAlex
- S2被引:511
- OpenAlex被引:3831
- 影响力被引:104
- TLDR:This work proposes a two time-scale update rule (TTUR) for training GANs with stochastic gradient descent on arbitrary GAN loss functions and introduces the "Frechet Inception Distance" (FID) which captures the similarity of generated images to real ones better than the Inception Score.
- OpenAlex ID:W4301206121
- OpenAlex DOI:10.48550/arxiv.1706.08500
- DOI:10.48550/arxiv.1706.08500
- DOI来源:OpenAlex
- 开放获取:green
- 开放获取链接:https://arxiv.org/pdf/1706.08500
- OpenAlex更新:2026-08-25
- 待LLM分类:否
- 成熟度:production
- 场景:GAN training、convergence、optimization
- 标题中文:使用双时间尺度更新规则训练的 GAN 收敛到局部纳什均衡
- TLDR中文:本文提出了一种双时间尺度更新规则(TTUR),用于在任意 GAN 损失函数下使用 SGD 训练 GAN,并引入了 Frechet Inception Distance(FID),相比 Inception Score 能更好地捕捉生成图像与真实图像之间的相似性。
- 来源文件:
- [OpenAlex discover]
- [S2 enrich]