计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (3): 185-188.

• 图形、图像、模式识别 • 上一篇    下一篇

基于分类贡献有效值的增量KNN模型修剪研究

周 靖,刘晋胜   

  1. 广东石油化工学院 计算机与电子信息学院,广东 茂名 525000
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2012-01-21 发布日期:2012-01-21

Research on incremental KNN model pruning based on classification contribution effective value

ZHOU Jing, LIU Jinsheng   

  1. College of Computer and Electronic Information, Guangdong University of Petrochemical Technology, Maoming, Guangdong 525000, China
  • Received:1900-01-01 Revised:1900-01-01 Online:2012-01-21 Published:2012-01-21

摘要: 增量学习的效果直接影响到KNN的效率和准确率。提出基于分类贡献有效值的增量KNN修剪模型(C2EV-KNNMODEL),将特征参数的分类贡献度与KNN增量学习结合起来,定义一种新的训练样本的贡献有效值,并根据此定义制定训练集模型的修剪策略。理论和实验表明,C2EV-KNNMODEL的适用性较强,能够使分类器的分类性能得到极大的提高。

关键词: K近邻分类, 分类贡献有效值, 增量学习

Abstract: The effect of incremental learning impacts on the efficiency and the rate of K-Nearest Neighbor algorithm directly. An incremental KNN model based on contribution effective value(CEV-KNNMODEL) is proposed, the paper combines the classification contribution degree and KNN incremental learning, defines a new contribution effective value of the training sample, and formulates the training set pruning strategy according to this definition. The theory and experiment shows that the applicability of CEV-KNNMODEL is strong, and the performance of the classifier can be greatly improved.

Key words: K-Nearest Neighbor(KNN), classification contribution effective value, incremental learning