BriefGPT.xyz
Jun, 2012
无投影在线学习
Projection-free Online Learning
HTML
PDF
Elad Hazan, Satyen Kale
TL;DR
本论文提出采用Frank-Wolfe技术的高效在线学习算法,避免了投影步骤,同时在在线凸优化方面获得了一系列遗憾界,特别是在随机在线平滑凸优化方面。此外,我们的算法具有参数自由性和产生稀疏决策的优点,并将算法应用到协作过滤的计算密集型应用中,并在标准数据集上显示出理论上的改进。
Abstract
The computational bottleneck in applying
online learning
to massive data sets is usually the projection step. We present efficient
online learning
algorithms that eschew projections in favor of much more efficien
→