arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.13605math.OC
收藏

用于(随机)非凸-凹极小极大问题的交替近端梯度步

Alternating proximal-gradient steps for (stochastic) nonconvex-concave minimax problems

Radu Ioan Boţ, Axel Böhm

AI总结:

本文针对非凸-凹极小极大问题,为随机交替梯度下降上升法证明了新的全局收敛速率,部分填补了该方法在非凸-凹设定下缺乏理论保证的空白。

摘要:

形如 $\min_x \max_y Ψ(x,y)$ 的极小极大问题由于机器学习,尤其是生成对抗网络的进展而受到越来越多的关注。这类问题通常使用针对两个参与者的随机梯度下降变体进行训练。尽管凸-凹问题已被充分理解,并且有许多高效的求解方法可供选择,但在该设定之外,即使是最简单的算法有时也缺乏理论保证。特别是交替梯度下降上升法,即两个智能体轮流更新其策略的情况正是如此。为了部分弥补文献中的这一空白,我们在非凸-凹设定下,针对寻找 $g(\cdot) := \max_y Ψ(\cdot,y)$ 临界点的该方法的随机版本,证明了一个新的全局收敛速率。

英文摘要:

Minimax problems of the form $\min_x \max_y Ψ(x,y)$ have attracted increased interest largely due to advances in machine learning, in particular generative adversarial networks. These are typically trained using variants of stochastic gradient descent for the two players. Although convex-concave problems are well understood with many efficient solution methods to choose from, theoretical guarantees outside of this setting are sometimes lacking even for the simplest algorithms. In particular, this is the case for alternating gradient descent ascent, where the two agents take turns updating their strategies. To partially close this gap in the literature we prove a novel global convergence rate for the stochastic version of this method for finding a critical point of $g(\cdot) := \max_y Ψ(\cdot,y)$ in a setting which is not convex-concave.

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。

↑