Discovering Anomalies on Mixed-Type Data Using a Generalized Student- t Based Approach

Anomaly detection in mixed-type data is an important problem that has not been well addressed in the machine learning field. Existing approaches focus on computational efficiency and their correlation modeling between mixed-type attributes is heuristically driven, lacking a statistical foundation. I...

Full description

Saved in:
Bibliographic Details
Published inIEEE transactions on knowledge and data engineering Vol. 28; no. 10; pp. 2582 - 2595
Main Authors Lu, Yen-Cheng, Chen, Feng, Wang, Yating, Lu, Chang-Tien
Format Journal Article
LanguageEnglish
Published IEEE 01.10.2016
Subjects
Anomaly detection
Bayes methods
Computational modeling
Correlation
Data models
Estimation
expectation propagation
mixed-type data
Numerical models
robust estimation
Robustness
variational inference
Online AccessGet full text

Cover

More Information
Summary:Anomaly detection in mixed-type data is an important problem that has not been well addressed in the machine learning field. Existing approaches focus on computational efficiency and their correlation modeling between mixed-type attributes is heuristically driven, lacking a statistical foundation. In this paper, we propose MIxed-Type Robust dEtection (MITRE), a robust error buffering approach for anomaly detection in mixed-type datasets. Because of its non-Gaussian design, the problem is analytically intractable. Two novel Bayesian inference approaches are utilized to solve the intractable inferences: Integrated-nested Laplace Approximation (INLA), and Expectation Propagation (EP) with Variational Expectation-Maximization (EM). A set of algorithmic optimizations is implemented to improve the computational efficiency. A comprehensive suite of experiments was conducted on both synthetic and real world data to test the effectiveness and efficiency of MITRE.
ISSN:1041-4347
1558-2191
DOI:10.1109/TKDE.2016.2583429