摘要
交叉覆盖算法分类时着重在于两类的交界部分,混杂在另外一类中往往无助于提高分类器的效率,反而会增加分类器的计算负担。本文提出一种基于交叉覆盖算法的最近邻交叉覆盖算法(NN-ACA):对进行训练的原始样本数据进行预处理,删除这些不同类的最近邻点,得到精简后的样本集,再对该样本集使用交叉覆盖算法。文章在介绍算法的同时,给出了相关实验数据,并对其和SVM进行了讨论,结果表明NN-ACA在一定的样本规模表现了速度和分类正确性上的优越性。
Alternative covering algorithm focuses on the samples near the boundary in training time , and those samples intermixed in another class are usually no good to improve the classifier＇s performance instead they may greatly increase the burden of computation. This paper gives a new algorithm-Nearest Neighbor Alternative Covering Algorithm based on Alternative covering algorithm（NN-ACA）. The new algorithms deal with data as following. Firstly, it preprocesses the samples by deleting the sample which has different class nearest neighbor. Secondly, the scaled samples are trained and tested by alternative covering algorithm. The result of experiment and discussion about the new algorithm and SVM are given by the paper. All of these are show that the new algorithm is better than alternative covering algorithm in speed and accuracy of classification of moderate size and dimension samples.
出处
《北京电子科技学院学报》
2007年第2期88-90,84,共4页
Journal of Beijing Electronic Science And Technology Institute
关键词
交叉覆盖
最近邻
分类
alternative covering
nearest neighbor
classification
