The application of extreme learning machines to the network intrusion detection problem

Abstract

The Extreme Learning Machine (ELM) algorithm conventionally suffers from the inferior batch training performance. In this paper, a new approach to combine ELM outputs is proposed with a view to further develop a persistent IDS. Specifically, this paper proposes the application of an Extreme Learning Machine based approach to the network-based intrusion detection system (IDSs). Good performance is achieved and preliminary results are reported in this paper.