Outlier Detection for Data Streams : Data Streams don't wait for processing

ISBN-13: 9783846591222 / Angielski / Miękka / 2011 / 200 str.

Shiblee Sadik

Opis

Outlier detection is a well established area of study for statistical data. However, most of the existing outlier detection techniques are designed targeting the regular data model, where the entire dataset is available for random access. Typical outlier detection techniques construct a standard data distribution or model from the entire dataset and execute their detection algorithms over each data point. Evidently these techniques are not suitable for online data streams where the entire dataset, due to its unbounded volume, is not available for random access. Moreover, the data distribution in data streams change over time which challenges the existing outlier detection techniques that assume a constant standard data distribution for the entire dataset. In addition, data streams are characterized by uncertainty which imposes further complexity. In this work we propose two outlier detection techniques, called Distance Based Outline Detection for Data Streams (DB-ODDS) and Automatic Outlier Detection for Data Streams (A-ODDS). Both techniques are based on a novel continuously adaptive data distribution function that addresses all the issues of data streams.

O autorze

Shiblee M. Sadik is a promising researcher is the area of data mining for data streams. He received his BSc degree in 2006 and MSc degree in 2010. He has been working with different research projects since 2005. He also worked as a professional software engineer for two years. He is pursuing his PhD degree now in the area of data stream mining.