文章基本信息

标题：Novel Approach of Usefulness of Reverse Neighbour Counts in Unsupervised Outlier Detection
本地全文：下载
作者：P. Munwar Gouse ; B.Nagalakshmi ; K.Ramesh 等
期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2017
卷号：6
期号：1
页码：1221
DOI：10.15680/IJIRSET.2017.0601131
出版社：S&S Publications
摘要：Outlier detection is an approach of finding outlying prototype from the given dataset. Outlier detectiongrew to become predominant field in specific talents domains. Information size is getting doubled each year there's ahave to observe outliers in huge datasets as early as viable. In high-dimensional knowledge outlier detection presentsquite a lot of challenges given that of curse of dimensionality. By using examining again the inspiration of reversenearest neighbors in the unsupervised outlier-detection context, high dimensionality can have a further have aninfluence on. In excessive dimensions it was once located that the distribution of facets in reverse-neighbor countsturns into skewed .This proposed work ambitions at developing and comparing one of the vital unsupervised outlierdetection ways and endorse a strategy to make stronger them. This proposed work goes in details in regards to thedevelopment and analysis of outlier detection algorithms such as neighborhood Outlier aspect (LOF), neighborhoodDistance-headquartered Outlier component(LDOF), Influenced Outliers and .The principles of those methods are thencombined to put in force a new approach with distributed strategy which improves the outcome of the prior recountedones on the subject of speed, complexity and accuracy.
关键词：Outlier detection; excessive-dimensional information; reverse nearest neighbors; unsupervised outlier;detection methods.