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 1906.04848cs.LGstat.ML
收藏

近距离观察生成对抗网络的优化景观

A Closer Look at the Optimization Landscapes of Generative Adversarial Networks

Mila, Université de Montréal(蒙特利尔大学Mila);Facebook AI Research(Facebook人工智能研究院);Element AI

Hugo Berard, Gauthier Gidel, Amjad Almahairi, Pascal Vincent, Simon Lacoste-Julien

摘要:

生成对抗网络在生成建模方面非常成功,但与标准深度神经网络相比,它们的训练仍然相对具有挑战性。在本文中,我们为GANs的优化景观提出了新的可视化技术,使我们能够研究由两个玩家的梯度级联产生的游戏向量场。使用这些可视化技术,我们试图弥合理论和实践之间的差距,通过经验表明,GANs的训练表现出围绕局部稳定点(LSSP)的显著旋转,类似于玩具示例的理论预测。此外,我们提供的经验证据表明,GAN训练收敛到一个稳定的稳定点,这是一个鞍点的发电机损失,而不是最小值,同时仍然实现了优异的性能。

英文摘要:

Generative adversarial networks have been very successful in generative modeling, however they remain relatively challenging to train compared to standard deep neural networks. In this paper, we propose new visualization techniques for the optimization landscapes of GANs that enable us to study the game vector field resulting from the concatenation of the gradient of both players. Using these visualization techniques we try to bridge the gap between theory and practice by showing empirically that the training of GANs exhibits significant rotations around Local Stable Stationary Points (LSSP), similar to the one predicted by theory on toy examples. Moreover, we provide empirical evidence that GAN training converge to a stable stationary point which is a saddle point for the generator loss, not a minimum, while still achieving excellent performance.

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。

↑