Differential games, in particular two-player sequential zero-sum games (a.k.a. minimax optimization), have been an important modeling tool in applied science and received renewed interest in machine learning due to many recent applications, such as adversarial training, generative models and reinforcement learning. However, existing theory mostly focuses on convex-concave functions with few exceptions. In this work, we propose two novel Newton-type algorithms for nonconvex-nonconcave minimax optimization. We prove their local convergence at strict local minimax points, which are surrogates of global solutions. We argue that our Newton-type algorithms nicely complement existing ones in that (a) they converge faster to strict local minimax points; (b) they are much more effective when the problem is ill-conditioned; (c) their computational complexity remains similar. We verify the effectiveness of our Newton-type algorithms through experiments on training GANs which are intrinsically nonconvex and ill-conditioned. Our code is available at https://github.com/watml/min-max-2nd-order.
翻译:不同的游戏,特别是两个玩家相继零和游戏(a.k.a.a. minimax优化),一直是应用科学中一个重要的模型工具,由于最近许多应用,例如对抗性训练、基因模型和强化学习等,对机器学习重新产生了兴趣;然而,现有的理论主要侧重于 convex concave 函数,只有少数例外。在这项工作中,我们为非convex-nonconconccave微型和最大优化提议了两种新型牛顿型算法。我们证明它们在严格的当地微型算法点的地方趋同,这些点是全球解决方案的代孕药。我们认为,我们的Newton型算法很好地补充了现有的算法,因为(a) 它们会更快地聚集到严格的当地微型算法点;(b) 当问题条件不成熟时,它们的效果要大得多;(c) 它们的计算复杂性仍然相似。我们通过培训本质上非convex和不完善的GANs实验来验证我们的Newton型算法的有效性。我们的代码可在 https://github.com/watml/min-max-toddservey.