The art of cyber threat hunting: harnessing AI for addressing newfangled cybersecurity challenges

Vista sencilla de ítem
dc.contributor.advisorPastor López, Iker
dc.contributor.advisorGarcía Bringas, Pablo
dc.contributor.authorMiranda García, Alberto
dc.date.accessioned2026-04-17T08:56:35Z
dc.date.available2026-04-17T08:56:35Z
dc.date.issued2025-04-04
dc.description.abstractThis thesis explores the integration of artificial intelligence into cybersecurity, addressing the growing complexity and dynamism of today's cyber threat landscape. As digital transformation speeds up in critical sectors, traditional static defence mechanisms can't keep up with today's adaptable and sophisticated attacks. This thesis argues that cybersecurity must evolve towards proactive, adaptive, and predictive defence models enabled by data-driven intelligence. The research is based on the hypothesis that artificial intelligence-based approaches significantly improve threat detection and response capabilities compared to conventional methods. To validate this, the study investigates four important cybersecurity challenges: spam and phishing detection, malware detection in portable executables, malware detection in Android, and anomaly detection in network traffic. For each domain, the thesis proposes and experimentally validates deep learning methodologies, such as LSTM networks for spam filtering, deep neural networks for malware classification in PE, and convolutional neural networks for bytecode analysis on Android and NetFlow traffic monitoring. Experimental results obtained with large-scale real and public datasets demonstrate a substantial increase in performance, with AI models achieving high accuracy, robustness, and generalisation ability. In particular, the transformation of Android bytecode into greyscale image representations for CNN-based analysis achieved detection accuracy of up to 99%, offering a novel and scalable approach to mobile security. Similarly, CNN-based NetFlow analysis validated the potential of AI in large-scale network monitoring.eng
dc.identifier.urihttps://hdl.handle.net/20.500.14454/5673
dc.language.isoeng
dc.publisherUniversidad de Deusto
dc.subjectMatemáticas
dc.subjectCiencia de los ordenadores
dc.subjectInteligencia artificial
dc.titleThe art of cyber threat hunting: harnessing AI for addressing newfangled cybersecurity challengeseng
dc.typedoctoral thesis
Archivos
Bloque original
Mostrando 1 - 1 de 1
Miniatura
Nombre:
miranda_art_2025.pdf
Tamaño:
4.76 MB
Formato:
Adobe Portable Document Format
Descargar
Colecciones
Tesis doctorales