A Scalable Hybrid Autoencoder–Extreme Learning Machine Framework for Adaptive Intrusion Detection in High-Dimensional Networks

The rapid expansion of network environments has introduced significant cybersecurity challenges, particularly in handling high-dimensional traffic and detecting sophisticated threats. This study presents a novel, scalable Hybrid Autoencoder–Extreme Learning Machine (AE–ELM) framework for Intrusion D...

Full description

Saved in:
Bibliographic Details
Published in:Future internet Vol. 17; no. 5; p. 221
Main Authors: Kumar, Anubhav, Radhakrishnan, Rajamani, Sumithra, Mani, Kaliyaperumal, Prabu, Balusamy, Balamurugan, Benedetto, Francesco
Format: Journal Article
Language:English
Published: Basel MDPI AG 01.05.2025
Subjects:
Accuracy
autoencoder
Classification
cloud continuum
Cybersecurity
Cyberterrorism
Datasets
Decision trees
Detectors
edge–fog–cloud orchestration
Efficiency
extreme learning machine
Feature selection
intrusion detection system
Intrusion detection systems
Machine learning
Optimization
Real time
scalable iot security
Support vector machines
ISSN:1999-5903, 1999-5903
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The rapid expansion of network environments has introduced significant cybersecurity challenges, particularly in handling high-dimensional traffic and detecting sophisticated threats. This study presents a novel, scalable Hybrid Autoencoder–Extreme Learning Machine (AE–ELM) framework for Intrusion Detection Systems (IDS), specifically designed to operate effectively in dynamic, cloud-supported IoT environments. The scientific novelty lies in the integration of an Autoencoder for deep feature compression with an Extreme Learning Machine for rapid and accurate classification, enhanced through adaptive thresholding techniques. Evaluated on the CSE-CIC-IDS2018 dataset, the proposed method demonstrates a high detection accuracy of 98.52%, outperforming conventional models in terms of precision, recall, and scalability. Additionally, the framework exhibits strong adaptability to emerging threats and reduced computational overhead, making it a practical solution for real-time, scalable IDS in next-generation network infrastructures.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1999-5903
1999-5903
DOI:10.3390/fi17050221