Localized Multiple Kernel learning for Anomaly Detection: One-class Classification

Abstract

Multi-kernel learning has been well explored in the recent past and has exhibited promising outcomes for multi-class classification and regression tasks. In this paper, we present a multiple kernel learning approach for the One-class Classification (OCC) task and employ it for anomaly detection. Recently, the basic multi-kernel approach has been proposed to solve the OCC problem, which is simply a convex combination of different kernels with equal weights. This paper proposes a Localized Multiple Kernel learning approach for Anomaly Detection (LMKAD) using OCC, where the weight for each kernel is assigned locally. Proposed LMKAD approach adapts the weight for each kernel using a gating function. The parameters of the gating function and one-class classifier are optimized simultaneously through a two-step optimization process. We present the empirical results of the performance of LMKAD on 25 benchmark datasets from various disciplines. This performance is evaluated against existing Multi Kernel Anomaly Detection (MKAD) algorithm, and four other existing kernel-based one-class classifiers to showcase the credibility of our approach. LMKAD achieves significantly better Gmean scores while using a lesser number of support vectors compared to MKAD. Friedman test is also performed to verify the statistical significance of the results claimed in this paper.

Anomaly detection for condition monitoring data using auxiliary feature vector and density‐based clustering

“…5 and Table 6 in [31]). The discharging faults include the low‐energy density arcing and arcing, whereas the overheating faults include the low and high temperature thermal. (iii) Semisupervised‐based method: LMKAD [7]. LMKAD is an extension of OCSVM.…”

“…Finally, the data of X (1) are inputted into the trained LMKAD model; the observations labelled with (−1) are anomalies. In this study, the best model LMKAD( S _ gpp ) in [7] is chosen for anomaly detection. (iv) Statistic‐based methods: EKPCA, Boxplot rule, and 2 σ rule of Gaussian distribution. EKPCA is a new PCA‐based method proposed in [16].…”

“…Finally, the data of X (1) are inputted into the trained LMKAD model; the observations labelled with (−1) are anomalies. In this study, the best model LMKAD( S _ gpp ) in [7] is chosen for anomaly detection.…”

“…In recent years, new and precise semisupervised methods have been proposed. Gautam et al [7] proposed a localised multiple kernel learning approach for anomaly detection (LMKAD), which outperformed conventional multiple kernel anomaly detection and OCSVM according to the experiment results from over 25 benchmark datasets. Miao et al [8] proposed a distributed OCSVM for wireless sensor network (WSN) anomaly detection, which requires relatively shorter running time and lower CPU memory consumption compared with OCSVM.…”

Combining Classifiers for Foreign Pattern Rejection

“…An interesting approach to anomaly detection based on a multiple kernel learning approach for the One-class Classification task is outlined in [19]. Localized Multiple Kernel learning approach for Anomaly Detection with One-class Classification method is proposed as an extension of known Multi Kernel Anomaly Detection, LMKAD, algorithm.…”