BriefGPT.xyz
Jun, 2015
使用灵活的小批量方案和非凸损失的ERM原始方法
Primal Method for ERM with Flexible Mini-batching Schemes and Non-convex Losses
HTML
PDF
Dominik Csiba, Peter Richtárik
TL;DR
本文提出了一种新算法用于规则经验风险最小化问题,可以使非凸损失函数也能收敛,达到与 QUARTZ 相同的复杂度,同时也能更好地利用数据信息,实现任意小批量的设计。
Abstract
In this work we develop a new
algorithm
for
regularized empirical risk minimization
. Our method extends recent techniques of Shalev-Shwartz [02/2015], which enable a dual-free analysis of
→