A Review of Methods for Detecting Point Anomalies on Numerical Dataset

Authors:

Abstract
Due to the fast development of anomaly detection techniques and its widely application fields, we attempt to provide a clear structure about the newly methods for detecting point anomaly in relative low dimensional numerical datasets. Firstly, we have reviewed recent advancements in three aspects including classification-based anomaly detection techniques, nearest-neighbor-based anomaly detection techniques and clustering-based anomaly detection techniques. Especially we make a detail survey about clustering-based anomaly detection techniques.

This publication has 40 references indexed in Scilit: