一种改进的应用于噪声数据中的KNN算法  

An Improvement KNN Algorithm To Deal with Noisy Data

作　　者：倪艾玲[1] 

机构地区：[1]安徽工业大学计算机学院,安徽马鞍山243002

出　　处：《计算机与现代化》2008年第7期75-78,共4页Computer and Modernization

摘　　要：基于实例的KNN算法不可避免地要依赖于数据的质量,但原始数据含有噪声,因而KNN算法的结果势必会因为数据中的噪声而受到严重的影响。事实上,大多噪声都服从一定的模型,而且模型一般是已知的。充分利用数据中的噪声模型,以减小噪声对KNN算法结果的影响。通过实验结果表明该方法是有效的。Case-based KNN algorithm inevitably has to rely on the quality of the data, but the original data is noisy. The result of KNN algorithm will inevitably be severely affected because of the noise data. In fact, most noise is subordinate to a certain model, and the model is generally known. Taking full advantage of the noise model in the data is to reduce the bad effect of noise on KNN algorithm results. Experimental results show that the method is effective.

关 键 词：分类 K-近邻算法 噪声模型 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]