Artificial intelligence, machine learning, and deep learning-based cybersecurity: Techniques, challenges, and future directions

Abstract

Sophisticated cyberattacks, ransomware, phishing, insider threats, and zero-day vulnerabilities have increased at an alarming rate, showing the weaknesses of traditional cybersecurity defenses, and an intelligent, adaptive, and automated cyber defense system is in high demand. Artificial Intelligence, Machine Learning, and Deep Learning have become transformative technologies in improving the cybersecurity resilience by identifying the threats in real-time, predicting the analytics, detecting anomalies, and automatically responding to the threat. The literature review is a systematic study of recent advances in Artificial Intelligence, Machine Learning, and Deep Learning-based cybersecurity methods that utilized PRISMA to get relevant studies in popular scientific databases, screen, and synthesize them based on the PRISMA paradigm. Cybersecurity systems such as intrusion detection systems, malware detection, phishing detection, ransomware detection, cyber threat intelligence, insider threat detection and cloud security are the areas of review. It additionally assesses the value of Explainable Artificial Intelligence, federated learning, generative AI, large language models, graph neural networks, reinforcement learning, and privacy-preserving machine learning in cyber defense in the present. Deep Neural Networks, Convolutional Neural Networks, Recurrent Neural Networks and hybrid AI architectures have demonstrated successful use in network security, internet of things security, endpoint detection and response and Security Information and Event Management settings. A conclusion of the review is that future ecosystems of cybersecurity will rely more on reliable AI, autonomous security systems, explainable models, and human-centered cyber resilience models to deal with the new cyber threats in the connected digital world.