Abstract:The ever-increasing interconnectedness of our world, fueled by technological advancements across industries, has made network security a paramount concern. This concern stems from the evolving tactics of cybercriminals and our dependence on interdependent systems. Extreme Learning Machines (ELMs) have recently been renamed Single-Hidden-Layer Feedforward Neural Networks (SLNFs) to achieve a rapid learning rate by randomly initializing weights and deviations. For a decade, researchers have been primarily focused on the investigations of ELM, which offers a distinctive use that prompted academics to investigate its possible application in various disciplines. Considering the Intrusion Detection System (IDS) framework, ELM provides an enticing path for constructing effective and adaptable IDS models capable of analyzing enormous amounts of network data in real-time. ELMs are based on feedforward networks using single or Multi-Hidden Layers (MHLs). ELM-based IDS assists in safeguarding networks, systems, and web applications against cyber-attacks, phishing, and cyber threats. As a result, during the last decade, ELM has taken center stage for development as an exciting technology for effective and accurate classification tasks. ELM represents a subset of the Machine Learning (ML) algorithms utilized by IDSs, which makes our work contribute significantly to the field of network security. First, we investigate the ELM-based IDS throughout the previous decade, including the idea, scope, and rationale of ELM. Second, our study aims to identify abnormal and malicious actions in network traffic while offering real-time protection against potential threats, including detecting Distributed Denial of Services (DDoS) attacks and phishing attacks that use malicious URLs and payloads for successful attacks. Finally, our work encompasses applications, extensions, attacks, security issues, and a comparative analysis showcasing the versatility of ELM within various algorithms. This underscores ELM's evolution as a promising technology for accurate and effective classification tasks, particularly in network security research.

A novel energy-based online sequential extreme learning machine to detect anomalies over real-time data streams

Anomaly Detection Algorithm Based on Electric Equipment

EADetection: an Efficient and Accurate Sequential Behavior Anomaly Detection Approach over Data Streams

Distributed Extreme Learning Machine for Nonlinear Learning over Network

An Extreme Learning Machine for Unsupervised Online Anomaly Detection in Multivariate Time Series

An Ensemble Extreme Learning Machine for Data Stream Classification

Prediction Based on Online Extreme Learning Machine in WWTP Application

An Algorithm of Robust Online Extreme Learning Machine for Dynamic Imbalanced Datasets

An Incremental Extreme Learning Machine for Online Sequential Learning Problems

Classifying Uncertain and Evolving Data Streams with Distributed Extreme Learning Machine

Online Network traffic anomaly detection method combining OS-ELM and SADE

ROS-ELM: A Robust Online Sequential Extreme Learning Machine for Big Data Analytics

Analysis of Extreme Learning Machines (ELMs) for intelligent intrusion detection systems: A survey

Anomaly Detection of Streaming Data Based on Deep Learning

Online sequential extreme learning machine with the increased classes

Online sequential extreme learning machine with forgetting mechanism.

Identifying Data Streams Anomalies by Evolving Spiking Restricted Boltzmann Machines.

Short Term Traffic Flow Prediction Based On On-Line Sequential Extreme Learning Machine

Regression-based Online Anomaly Detection for Smart Grid Data

A User Behavior Anomaly Detection Approach Based on Sequence Mining over Data Streams.

A Novel Online Sequential Learning Algorithm for ELM Based on Optimal Control