摘要: 任何涉及k近邻求解问题的算法被应用于处理不同特征的数据集时,参数k值的选择都会明显影响算法的性能和结果。因而,如何选择k近邻算法中敏感参数k值一直是一个研究难点。提出了一种新的近邻关系——自然最近邻,它不需要设置参数k,每个节点的邻居是由算法自适应计算而形成的。针对离群点检测的特殊性,通过确定自然最近邻居搜索算法的终止条件,提出一种基于自然最近邻的新的离群检测算法ODb3N。实验表明,该算法不仅避免了k近邻中参数的选择问题,而且能够更有效地发现离群簇。
[1] Gogoi P,Borah B,Bhattacharyya D K.Outlier identificationusing symmetric neighborhoods[J].Procedia Technology,2012,6:239-246 [2] Breunig M M,Kriegel H P,et al.LOF:identifying density-based local outliers[J].Proc.of 2000ACM SIGMOD international conference on Management of data.ACM Sigmod Record,2000,29(2):93-104 (下转第305页)(上接第278页) [3] Hautamaki V,Karkkainen I.Outlier detection using k-nearestneighbor graph[C]∥Proc.17th IEEE Int.Conf.on Pattern Recognition.2004,3:430-433 [4] Angiulli,F,Palopoli L.Detecting outlying properties of exceptional objects[J].ACM Transaction on Database Systems,2009,34(1):62-74 [5] Richard J,Chris C.Fuzzy-rough nearest neighbor classificationand prediction[J].Theoretical Computer Science,2011,412(42):5871-5884 [6] Pandya D H,Upadhyay S H,Harsha S P.Fault diagnosis of rol-ling element bearing with intrinsic mode function of acoustic emission data using APF-kNN[J].Expert Systems with Applications,2013,40(10):4137-4145 [7] Xu Yong,Zhu Qi,et al.Coarse to fine K nearest neighbor classifier [J].Pattern Recognition Letters,2013,34(9):980-986 [8] 符永铨,王意洁.DKNNS:面向延迟敏感型应用的可扩展精确分布式K近邻搜索算法研究[J].中国科学(信息科学),2012,42(5):561-577 |
No related articles found! |
|