Accelerated Methods for $\alpha$-Weakly-Quasi-Convex Problems
Accelerated Methods for $\alpha$-Weakly-Quasi-Convex Problems
复制标题
$alpha$-弱拟凸问题的加速方法
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
A. Gasnikov
中科院分区：
文献类型：
--
作者：
Sergey Guminov;A. Gasnikov
训练神经网络中遇到的许多问题都是非凸。但是，其中一些人满足条件比凸度弱，但仍足以保证某些一阶方法的收敛性。在我们的工作中，我们表明一些以前已知的一阶方法在这些较弱的条件下保留其收敛速率。
Many problems encountered in training neural networks are non-convex. However, some of them satisfy conditions weaker than convexity, but which are still sufficient to guarantee the convergence of some first-order methods. In our work we show that some previously known first-order methods retain their convergence rates under these weaker conditions.