Attack Detection in Mobile Internet and Networks Using the Graph-Based Schemes for Combining the Support Vector Machines

• Published in: Mobile Internet Security Vol. 971; pp. 1 - 16
• Main Authors: Branitskiy, Alexander, Kotenko, Igor
• Format: Book Chapter
• Language: English
• Published: Singapore Springer Singapore Pte. Limited 2018; Springer Singapore
• Series: Communications in Computer and Information Science
• Subjects: Classification binary tree; Directed acyclic graph; Network attack detection; Principal component analysis; Support vector machine
• ISBN: 9789811337314; 9811337314
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-981-13-3732-1_1

The paper presents a comparative analysis of two schemes for combining the binary classifiers. In the role of such classifiers we use well-known models—support vector machines (SVMs). For constructing the multiclass models we experimentally investigate two schemes for combining the SVMs, namely a classification binary tree (CBT) and a directed acyclic graph (DAG). Main application of considered models we demonstrate in the paper is attack detection and classification in mobile Internet and networks. The various performance indicators of classifiers are given. The results of experiments performed for to estimate these indicators and usage of time and system resources are presented.