A comparative study of ML-ELM and DNN for intrusion detection
Abstract
© 2021 ACM. Intrusion detection remains one of the critical research issues in network security. Many machine learning algorithms have been proposed to develop intrusion detection systems, which can categorize network traffic into normal and anomalous classes. The multilayer extreme learning machine (ML-ELM) and the deep neural network (DNN) are two machine learning algorithms based on different theories/concepts that use the same multilayer architecture. In this paper, a comparative study is performed to shed light on the selection between these two algorithms with the same architecture in intrusion detection applications. The study explores the performance of the ML-ELM and DNN algorithms under similar parameter settings. With in-depth analysis and discussions, the limitations and advantages of each algorithm are outlined. In addition, the performance trend of each algorithm with increasing parameter values is studied.
RAS ID
32899
Document Type
Conference Proceeding
Date of Publication
2021
Funding Information
Cyber Security Research Centre Limited Australian Government Cooperative Research Centres (CRC) Program
School
School of Science / ECU Security Research Institute
Copyright
subscription content
Publisher
Association for Computing Machinery
Comments
Yang, W., Wang, S., & Johnstone, M. (2021, February). A comparative study of ML-ELM and DNN for intrusion detection [Paper presentation]. ACSW '21: 2021 Australasian Computer Science Week Multiconference, Dunedin, New Zealand. https://doi.org/10.1145/3437378.3437390