2018/05/21 by Chandan Gautam, Gautam, Chandan, Ramesh Balaji +7 · 1 citation
Computer Science · Engineering · #Advanced Chemical Sensor Technologies #Anomaly Detection Techniques and Applications #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Network Security and Intrusion Detection
paper · pdf · doi:10.48550/arxiv.1805.07892
openalex publication_date 2018/05/21 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
Multi-kernel learning has been well explored in the recent past and has\nexhibited promising outcomes for multi-class classification and regression\ntasks. In this paper, we present a multiple kernel learning approach for the\nOne-class Classification (OCC) task and employ it for anomaly detection.\nRecently, the basic multi-kernel approach has been proposed to solve the OCC\nproblem, which is simply a convex combination of different kernels with equal\nweights. This paper proposes a Localized Multiple Kernel learning approach for\nAnomaly Detection (LMKAD) using OCC, where the weight for each kernel is\nassigned locally. Proposed LMKAD approach adapts the weight for each kernel\nusing a gating function. The parameters of the gating function and one-class\nclassifier are optimized simultaneously through a two-step optimization\nprocess. We present the empirical results of the performance of LMKAD on 25\nbenchmark datasets from various disciplines. This performance is evaluated\nagainst existing Multi Kernel Anomaly Detection (MKAD) algorithm, and four\nother existing kernel-based one-class classifiers to showcase the credibility\nof our approach. Our algorithm achieves significantly better Gmean scores while\nusing a lesser number of support vectors compared to MKAD. Friedman test is\nalso performed to verify the statistical significance of the results claimed in\nthis paper.\n