BriefGPT.xyz
Feb, 2019
大规模多类别分类的高效原始 - 对偶算法
Efficient Primal-Dual Algorithms for Large-Scale Multiclass Classification
HTML
PDF
Dmitry Babichev, Dmitrii Ostrovskii, Francis Bach
TL;DR
本文介绍了如何使用随机镜像下降法和非均匀采样方案，来快速训练高维度特征空间、多分类通用的线性分类器，特别是在多类Hinge损失下，本文提出了一个迭代次数为$O(d+n+k)$的子线性算法。
Abstract
We develop efficient algorithms to train $\ell_1$-regularized
linear classifiers
with large
dimensionality
$d$ of the feature space, number of classes $k$, and sample size $n$. Our focus is on a special class of
→