摘要
严格有序分支的快速最近邻算法保证节点距离和样本距离的严格性,能以约4倍于NN法则的速度取得与其完全相同的分类正确率,从而在不损失正确性的前提下提高了快速性。在其基础上修改得到的综合算法,则以远小于NN法则的空间和时间开销取得了相当接近于NN法则的分类结果。
Based on the strict ordered partition, a fast K-NN algorithm is presented which ensures the strictness of the distance from the test sample to the nodes or the training samples. It is able to take the same right-rate of classification four times as fast as the NN-rule does. The composite algorithm is obtained with some modifications to K-NN algorithm. With far small cost of space and time, it could achieve an approaching affect of classification to the NN-rule.
出处
《计算机学报》
EI
CSCD
北大核心
1990年第7期489-497,共9页
Chinese Journal of Computers
