Deep Learning Methods to Prevent Various Cyberattacks in Cloud Environment
Qusay Kanaan Kadhim, Ohood Fadil Alwan, Inteasar Yaseen Khudhair · Revue d intelligence artificielle · 2024
Cloud computing offers many benefits, but it also presents new cybersecurity challenges that must be addressed to ensure data protection in the cloud environment.Governments and organizations face increasing and ongoing cyberattacks by state-sponsored hackers to wage cyberwars.A successful cyberattack on vital infrastructure, such communications or electricity networks, might have disastrous effects.These attacks vary in their forms and patterns, which makes understanding and confronting them necessary.An essential function of artificial intelligence is the detection of intrusions.To prevent various cyberattacks in the cloud environment and is widely considered the best method.The Deep Learning (DL) method efficiently trained on datasets to improve performance based on statistical features can accurately detect various attacks.In this paper, we use the CE-CIC-2018 dataset that contains seven distinct attack scenarios, updated for cybersecurity: Brute-force, Heartbleed, Botnet, DoS, DDoS, and Web Attacks.This paper contributes, to improving the precision of identifying different types of threats in a cloud environment and improving additional performance indicators.The proposed using DL method dimensionality reduction using Principal Component Analysis (PCA), the Fuzzy C-Means (FCM) technique to create clusters, and the deep learning-based AutoEncoder (AE) method combined to identify the attack and non-attack.PCA + FCM + AE method prevents different cyberattacks in a cloud environment.The results showed that the best accuracy was 97.70 %, which is the highest accuracy compared to those results reported in the relevant literature.