arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2103.12685cs.LGcs.AI
收藏

生成最小化网络:无竞争训练GANs

Generative Minimization Networks: Training GANs Without Competition

ETH Zurich(苏黎世联邦理工学院);Technion-Israel Institute of Technology(以色列理工学院)

Paulina Grnarova, Yannic Kilcher, Kfir Y. Levy, Aurelien Lucchi, Thomas Hofmann

摘要:

机器学习中的许多应用可以被构造为最小化问题,并使用基于梯度的技术有效地解决。然而,最近生成模型的应用,特别是GANs,引发了人们对解决标准优化技术通常不适用的最小-最大游戏的兴趣。从业者经历的已知问题之一是缺乏收敛保证或收敛到非最优循环。这些问题的核心是GAN目标的最小-最大结构,它在参与者之间创建了非平凡的依赖关系。我们建议通过优化一个不同的目标来解决这个问题,该目标使用博弈论中的二元差距概念来规避最小-最大结构。我们提供了新的收敛保证在这个目标上,并证明了为什么得到的极限点解决问题比已知的技术。

英文摘要:

Many applications in machine learning can be framed as minimization problems and solved efficiently using gradient-based techniques. However, recent applications of generative models, particularly GANs, have triggered interest in solving min-max games for which standard optimization techniques are often not suitable. Among known problems experienced by practitioners is the lack of convergence guarantees or convergence to a non-optimum cycle. At the heart of these problems is the min-max structure of the GAN objective which creates non-trivial dependencies between the players. We propose to address this problem by optimizing a different objective that circumvents the min-max structure using the notion of duality gap from game theory. We provide novel convergence guarantees on this objective and demonstrate why the obtained limit point solves the problem better than known techniques.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑