Development of an Intrusion Detection System Using Multilayer Perceptron

EasyChair Preprint 15530

Abstract

This abstract explores the development of an Intrusion Detection System (IDS) leveraging a Multilayer Perceptron (MLP) neural network to enhance network security by detecting malicious activity. IDSs are essential tools for identifying unauthorized access and potential cyber threats within a network. Traditional IDS methods, such as signature-based and anomaly-based detection, often struggle to recognize novel and complex attacks. MLP-based IDSs address this limitation by learning intricate patterns in network traffic through multiple layers of neurons, enabling the detection of both known and unknown intrusions. This development process involves data collection, preprocessing, and feature selection, followed by designing and training the MLP model using labeled network traffic data. Model evaluation is conducted through accuracy, precision, recall, and F1 score metrics to ensure reliable performance. While challenges such as data quality, computational demands, and model interpretability remain, MLP-based IDSs show significant promise in advancing network protection. The deployment of such systems contributes to adaptive, real-time detection capabilities, improving resilience against evolving cyber threats.

Keyphrases: Artificial Neural Network, Cybersecurity, Intrusion Detection System, Multilayer Perceptron, Network Security

@booklet{EasyChair:15530,
  author    = {Israel Echeta},
  title     = {Development of an Intrusion Detection System Using Multilayer Perceptron},
  howpublished = {EasyChair Preprint 15530},
  year      = {EasyChair, 2024}}