AI-Enhanced Threat Detection and Response in Cloud Infrastructure Using Deep Learning Techniques

Abstract

Cybersecurity must be proactive and flexible to keep up with the ever-changing cyber threat scenario. As cloud infrastructure becomes a primary target for cyber-attacks, there is a growing need for advanced threat detection systems to protect critical data and services. This document adopts deep learning methods to establish an AI-based threat detection structure for protecting cloud security. The framework applies the Edge-IIoT set dataset to detect numerous attack scenarios that occur in IoT and IIoT environments, particularly DoS/DDoS attacks. The framework uses Edge-IIoTset dataset which goes through strict data preprocessing steps and normalization and feature selection to enhance model optimization. The obtained experimental findings establish BERT as the top model because it delivers a detection accuracy of 98.2% and precision/recall/F1-score readings of 98%, which validate its superior ability to detect
complex threat patterns. The BERT model achieves high classification accuracy across attack types based on its performance analyses through confusion  matrix and ROC curve. The study demonstrates how BERT contributes to enhancing cloud infrastructure cybersecurity frameworks by delivering dependable and scalable solutions for protecting Industrial Internet of Things networks from threats.