Outlier detection based on approximation accuracy entropy

Feng Jiang

doi:10.1007/s13042-018-0884-8

Outlier detection based on approximation accuracy entropy

2018-11-20

25

PRE

AI

DOI：10.1007/s13042-018-0884-8

Original

Original request for help

Save

Abstract

En 中文

Recently, proximity-based outlier detection methods receive much attention. For any given object x, a proximity-based method usually measures the degree of outlierness of x through examining the nearest neighbor structure of x, where the size of nearest neighborhood should be predetermined by the users. However, it is difficult for users to determine the size of nearest neighborhood. To solve the above problem, in this paper, we present an approximation accuracy entropy-based outlier detection algorithm, called ODAAE, within the framework of rough sets. Approximation accuracy entropy is an extension of Shannon information entropy in rough sets. To quantify the degree of outlierness of any given object, we develop a measure called the AAE(approximation accuracy entropy)-based outlier factor. Experimental results on real-world data sets show that the proposed algorithm is effective for outlier detection.

Keywords:

Outlier detection

Rough sets

Approximation accuracy

Conditional entropy

Approximation accuracy entropy

Outlier detection based on approximation accuracy entropy

Abstract

AI Summary

Journal

Organization