The increasing sophistication and frequency of cyber threats in today's interconnected digital world, traditional security mechanisms in timely and accurate threat detection. The both data volume and attack surface growing exponentially, modern cyber security needs intelligent, adaptive, and scalable solutions. This paper was investigating AI and ML techniques for cyber threat prediction and prevention. The major assesses to AI-based threat detection models increase the accuracy, speed, and automation of threat detection systems, anticipate and work against security incidents. The proposed work of existing AI and ML techniques used in cybersecurity, including supervised and unsupervised learning, deep learning, and anomaly detection techniques. These methods have been assessed in various areas such as the Intrusion Detection Systems (IDS), behaviour-based threat prediction, and automated response mechanisms to security incidents. The study of crucial issues such as adversarial attacks, data bias, and model interpretability.
30 September 2026 in Artificial Intelligence and Emerging Technologies
Cloud Computing has emerged as an essential platform for the modern applications. But the growing complexity of the attacks puts pressure on the security of cloud environments. Current intrusion detection systems (IDS) have weaknesses of detection inaccuracy, high false alarm rate and low efficiency in processing high-dimensional network traffic data. To overcome these drawbacks, this study suggests an intelligent intrusion detection framework based on XGBoost algorithm with correlation-based feature selection to remove redundant features, decrease the computational complexity, and enhance the classification performance. The proposed framework is based on UNSW-NB15 dataset, which consists of realistic normal and malicious network traffic, and data preprocessing methods such as categorical encoding, feature scaling, and a 70:30 ratio for training and testing. The results of this experimental study show that the proposed model attains 99.11% accuracy, 99.13% precision, 99.57% recall, 99.35% F1 score and 99.84% AUC which is higher than that achieved by the traditional machine learning models like random forest, decision tree, naïve bayes etc. The results show that the proposed framework is accurate, efficient and scalable solution for intelligent cloud intrusion detection which improves the security and reliability of cloud computing environments.