BriefGPT.xyz
May, 2010
通过无标签实例的选择来改善半监督支持向量机
S4VM: Safe Semi-Supervised Support Vector Machine
HTML
PDF
Yu-Feng Li, Zhi-Hua Zhou
TL;DR
本文提出了一种 S3VM-us 方法，使用分层聚类选择无标签实例，以减少 S3VMs 的性能退化机会，在广泛的数据集上进行实验表明，S3VM-us 的性能退化机会远小于现有的 S3VMs。
Abstract
Semi-supervised learning tries to improve performance by using
unlabeled data
. In some situations, however, its performance may become inferior to that of without using
unlabeled data
. It is desired to have safe
→