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.

+More

anomaly detection in mixedtype machine learning computational efficiency correlation modeling between mixedtype attributes expectation propagation ep robust error buffering approach laplace approximation inla robust detection mitre inferences bayesian inference approaches variational expectationmaximization synthetic and real world data statistical foundation

Yating Wang, Chang-Tien Lu,Yen-Cheng Lu, Feng Chen,.Discovering Anomalies on Mixed-Type Data Using a Generalized Student- Based Approach. 28 (10),2582-2595.

Yating Wang, et al."Discovering Anomalies on Mixed-Type Data Using a Generalized Student- Based Approach" 28

Yating Wang, Chang-Tien Lu,Yen-Cheng Lu, Feng Chen,"Discovering Anomalies on Mixed-Type Data Using a Generalized Student- Based Approach"