arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2103.15985math.OCcs.NE
收藏

鞍点Optimization与近似最小化Oracle

Saddle Point Optimization with Approximate Minimization Oracle

Youhei Akimoto

摘要:

鞍点优化$\min_x\max_y f(x, y)$的一种主要方法是基于梯度的方法,如生成对抗网络(GANs)所推广的。相比之下,我们分析了一种替代方法,只依赖于一个甲骨文,解决了一个最小化问题近似。我们的方法将近似解$x'$和$y'$定位到给定点$(x, y)$处的$\min_{x'}f(x', y)$和$\max_{y'}f(x, y')$,并以学习率$η$向这些近似解$(x', y')$更新$(x, y)$。在局部强凸-凹光滑函数上,我们得到了$η$线性收敛于局部鞍点的条件,这揭示了最近发展的鲁棒对抗强化学习算法的一个可能的缺点.我们开发了一个启发式的方法来适应$η$导数自由和实现零阶和一阶最小化算法。数值实验表明了理论结果的严密性以及$η$自适应机制的有效性。

英文摘要:

A major approach to saddle point optimization $\min_x\max_y f(x, y)$ is a gradient based approach as is popularized by generative adversarial networks (GANs). In contrast, we analyze an alternative approach relying only on an oracle that solves a minimization problem approximately. Our approach locates approximate solutions $x'$ and $y'$ to $\min_{x'}f(x', y)$ and $\max_{y'}f(x, y')$ at a given point $(x, y)$ and updates $(x, y)$ toward these approximate solutions $(x', y')$ with a learning rate $η$. On locally strong convex--concave smooth functions, we derive conditions on $η$ to exhibit linear convergence to a local saddle point, which reveals a possible shortcoming of recently developed robust adversarial reinforcement learning algorithms. We develop a heuristic approach to adapt $η$ derivative-free and implement zero-order and first-order minimization algorithms. Numerical experiments are conducted to show the tightness of the theoretical results as well as the usefulness of the $η$ adaptation mechanism.

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。

↑