Foundations of Deep Learning and Large Language Models in Cybersecurity DOI
Hewa Majeed Zangana, Marwan Omar, Jamal N. Al-Karaki

и другие.

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 1 - 36

Опубликована: Апрель 8, 2025

The integration of deep learning (DL) and large language models (LLMs) has significantly advanced the field cybersecurity, offering innovative approaches to threat detection, anomaly identification, secure communication. Deep techniques, such as neural networks reinforcement learning, have demonstrated robust capabilities in detecting previously unknown threats by patterns from vast amounts cybersecurity data. Similarly, LLMs, particularly transformers, revolutionized natural processing tasks, enabling effective vulnerability analysis, malware classification, phishing detection. This chapter explores foundational concepts highlighting their applications challenges within landscape. Additionally, it discusses synergy between these technologies, focusing on how they complement traditional measures drive evolution intelligent defense mechanisms.

Язык: Английский

Secure Authentication and Identity Management With AI DOI

Derek Mohammed,

Helen MacLennan

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 271 - 306

Опубликована: Апрель 8, 2025

Secure authentication and identity management are critical components of modern cybersecurity, ensuring that only authorized users gain access to sensitive systems data. Traditional methods, such as passwords multi-factor authentication, face increasing challenges due sophisticated cyber threats, credential theft, user experience limitations. Recent advancements in artificial intelligence (AI), particularly deep learning large language models (LLMs), have revolutionized mechanisms by enhancing security, accuracy, adaptability. AI-driven leverage biometric recognition, behavioral analysis, anomaly detection improve verification fraud prevention. Additionally, federated decentralized frameworks provide robust solutions for privacy-preserving authentication. This chapter explores the integration AI secure management, discussing its benefits, challenges, future research directions.

Язык: Английский

Процитировано

0
AI Automated Incident Response and Threat Mitigation Using AI DOI
Rebet Jones

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 201 - 236

Опубликована: Апрель 8, 2025

The rapid evolution of cyber threats has necessitated the development advanced techniques for incident response and threat mitigation. Artificial Intelligence (AI) emerged as a transformative force in cybersecurity, particularly automating detection, response, mitigation processes. This chapter explores role AI, including Deep Learning (DL) Large Language Models (LLMs), revolutionizing strategies. By leveraging organizations can achieve faster, more accurate adaptive mechanisms, efficient strategies, significantly improving their overall security posture. examines key AI technologies, applications challenges faced, future potential AI-driven operations.

Язык: Английский

Процитировано

0
Malware Analysis and Classification Using Deep Learning DOI
Rebet Jones

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 165 - 200

Опубликована: Апрель 8, 2025

Malware analysis and classification have become critical components of modern cybersecurity strategies, given the increasing sophistication cyber threats. With rapid advancement machine learning techniques, particularly deep learning, ability to detect classify malware has improved significantly. This chapter explores role in automating detection classification, focusing on use neural networks, feature extraction, pattern recognition. We discuss various architectures, such as Convolutional Neural Networks (CNNs) Recurrent (RNNs), that been successfully applied analysis. Additionally, examines challenges limitations using models, including data imbalance, overfitting, model interpretability. The integration large datasets, along with potential language models (LLMs), is also explored for enhanced accuracy.

Язык: Английский

Процитировано

0
Phishing and Social Engineering Attack Prevention With LLMs DOI
Hermano Jorge Da Silva De Queiroz

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 133 - 164

Опубликована: Апрель 8, 2025

Phishing and social engineering attacks have become increasingly sophisticated, leveraging advanced psychological manipulation deceptive tactics to compromise individuals organizations. Traditional cybersecurity measures often fail address the dynamic evolving nature of these threats. Large Language Models (LLMs) present a promising solution by enabling real-time threat detection, automated phishing email classification, user education through natural language processing capabilities. This chapter explores how LLMs can enhance attack prevention, discussing their role in filtering, anomaly chatbot-based awareness training, behavioral analysis. Additionally, we challenges such as adversarial LLMs, ethical considerations, model biases. By integrating deep learning with frameworks, organizations develop more resilient adaptive defense mechanisms against human-targeted cyber

Язык: Английский

Процитировано

0
Threat Detection and Anomaly Identification Using Deep Learning DOI
Hermano Jorge Da Silva De Queiroz, Helen MacLennan

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 65 - 96

Опубликована: Апрель 8, 2025

Cyber threats are increasingly complex, requiring advanced detection and mitigation strategies. Deep learning (DL) offers powerful solutions for threat anomaly identification, thanks to its ability process large data volumes uncover subtle indicators of cyber threats. This chapter discusses the integration various DL techniques—CNNs, RNNs, autoencoders, GANs—for detecting malicious activities anomalies across diverse cybersecurity contexts. By examining both supervised unsupervised approaches, it highlights strengths limitations in tackling such as zero-day attacks, insider threats, APTs. Real-world applications, case studies, role explainable AI (XAI) enhancing transparency trust also explored. Finally, challenges like adversarial quality, computational constraints addressed, along with future directions improving robustness efficiency DL-driven systems.

Язык: Английский

Процитировано

0
Foundations of Deep Learning and Large Language Models in Cybersecurity DOI
Hewa Majeed Zangana, Marwan Omar, Jamal N. Al-Karaki

и другие.

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 1 - 36

Опубликована: Апрель 8, 2025

The integration of deep learning (DL) and large language models (LLMs) has significantly advanced the field cybersecurity, offering innovative approaches to threat detection, anomaly identification, secure communication. Deep techniques, such as neural networks reinforcement learning, have demonstrated robust capabilities in detecting previously unknown threats by patterns from vast amounts cybersecurity data. Similarly, LLMs, particularly transformers, revolutionized natural processing tasks, enabling effective vulnerability analysis, malware classification, phishing detection. This chapter explores foundational concepts highlighting their applications challenges within landscape. Additionally, it discusses synergy between these technologies, focusing on how they complement traditional measures drive evolution intelligent defense mechanisms.

Язык: Английский

Процитировано

0