计算机应用与软件2011，Vol.28Issue(9)：229-231,3.
基于密度敏感距离的协同训练算法
DENSITY-SENSITIVE SEMI-SUPERVISED CO-TRAINING
徐飞裕 1徐荣聪1
作者信息
- 1. 福州大学数学与计算机科学学院 福建福州350108
- 折叠
摘要
Abstract
Semi-supervised algorithm, as an algorithm effectively exploiting large amounts of unlabeled data to improve the capability of classifier trained by small amount of labelled data, has its significance no matter in theory or in practice. A novel co-training algorithm based on density-sensitive distance is proposed in the paper. The algorithm introduces a distance metric which is able to depict inherent clustering distribution of data well and in this way to boost the performance of weak classifier derived by combined classifier on small amounts of dataset. The method is indicated effective through experiments.关键词
密度敏感/协同训练/分类器组合/半监督学习Key words
Density-sensitive Co-training Classifier combination Semi-supervised learning分类
信息技术与安全科学引用本文复制引用
徐飞裕,徐荣聪..基于密度敏感距离的协同训练算法[J].计算机应用与软件,2011,28(9):229-231,3.
