Rapid Forecasting of Cyber Events Using Machine Learning-Enabled Features

Ahmed, Yussuf and Azad, Muhammad Ajmal and Asyhari, Taufiq (2024) Rapid Forecasting of Cyber Events Using Machine Learning-Enabled Features. Information, 15 (1). p. 36. ISSN 2078-2489

[thumbnail of information-15-00036.pdf]
Preview
Text
information-15-00036.pdf - Published Version
Available under License Creative Commons Attribution.

Download (5MB)

Abstract

In recent years, there has been a notable surge in both the complexity and volume of targeted cyber attacks, largely due to heightened vulnerabilities in widely adopted technologies. The Prediction and detection of early attacks are vital to mitigating potential risks from cyber attacks and network resilience. With the rapid increase of digital data and the increasing complexity of cyber attacks, big data has become a crucial tool for intrusion detection and forecasting. By leveraging the capabilities of unstructured big data, intrusion detection and forecasting systems can become more effective in detecting and preventing cyber attacks and anomalies. While some progress has been made on attack prediction, little attention has been given to forecasting cyber events based on time series and unstructured big data. In this research, we used the CSE-CIC-IDS2018 dataset, a comprehensive dataset containing several attacks on a realistic network. Then we used time-series forecasting techniques to construct time-series models with tuned parameters to assess the effectiveness of these techniques, which include Sequential Minimal Optimisation for regression (SMOreg), linear regression and Long Short-Term Memory (LSTM) to forecast the cyber events. We used machine learning algorithms such as Naive Bayes and random forest to evaluate the performance of the models. The best performance results of 90.4% were achieved with Support Vector Machine (SVM) and random forest. Additionally, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) metrics were used to evaluate forecasted event performance. SMOreg’s forecasted events yielded the lowest MAE, while those from linear regression exhibited the lowest RMSE. This work is anticipated to contribute to effective cyber threat detection, aiming to reduce security breaches within critical infrastructure.

Item Type: Article
Identification Number: 10.3390/info15010036
Dates:
Date
Event
5 January 2024
Accepted
2024
Published Online
Uncontrolled Keywords: forecasting, big data, time series, cyber attack prediction, cyber events, intrusion detection
Subjects: CAH11 - computing > CAH11-01 - computing > CAH11-01-01 - computer science
Divisions: Faculty of Computing, Engineering and the Built Environment > College of Computing
Depositing User: Gemma Tonks
Date Deposited: 09 May 2024 14:02
Last Modified: 09 May 2024 14:02
URI: https://www.open-access.bcu.ac.uk/id/eprint/15482

Actions (login required)

View Item View Item

Research

In this section...