A Meta-Heuristic Online Support Vector Machines for Big Data Cyber-Security

G.A. Mylavathi and Dr.B. Srinivasan

Recently, Cyber security is observed to be a serious problem in big data and it presents a very big challenge to the research field. Machine learning techniques had enforced on the particular candidates, especially to deal with the problems of big data security. Among all of the algorithms, the support vector machines have been remarkably successful on various classification problems. For the establishment of a highly efficient Support Vector Machines (SVMs), user has to define the right SVM configuration in prior and this is one challenge that demands the professional knowledge and huge amount of the human effort for the trial and error task. The important objective of this research work is to design a novel classifier that makes use performance for managing the specific choice of low-level heuristic. In this research, the Online Support Vector Machines (OSVMs) configuration process is formulated as the bi-objective optimization problem in which accuracy and the model complexity are regarded to be the two contrasting goals. This review presents a hyper meta-heuristic framework using artificial bee colony (ABC) for a bi-objective optimization that is independent of the problem domain. The low level-heuristic aids in the generation of new OSVMs configuration for issues involving cyber security. The results achieved could show that the new framework has been highly efficient.

Volume 11 | 01-Special Issue

Pages: 87-95