OntoLogX: Ontology-Guided Knowledge Graph Extraction From Cybersecurity Logs With Large Language Models
Published Wiley version of OntoLogX, which turns raw logs into ontology-grounded knowledge graphs for actionable cyber threat intelligence.
Publications, technical reports, and funding
Physics joins AI: A real-time hybrid misbehavior detection framework for vehicular networks
Paper on a hybrid system that combines AI analysis and physical coherence checks to detect anomalous behaviour in vehicular networks and improve cooperative driving safety.
OntoLogX: Ontology-Guided Knowledge Graph Extraction from Cybersecurity Logs with Large Language Models
Technical report on ontology-guided knowledge-graph extraction from cybersecurity logs using LLMs.
Enabling Transparent Cyber Threat Intelligence Combining Large Language Models and Domain Ontologies
Technical report on transparent CTI that combines LLMs and domain ontologies to improve structure, traceability, and explainability.
Preliminary Investigation into Uncertainty-Aware Attack Stage Classification
Preliminary technical report on classifying attack stages while explicitly modelling uncertainty.
TRUE
Project on interpretable and trustworthy situational understanding for multivariate time series, combining spatio-temporal learning, argumentation-based causal inference, and uncertainty-aware risk assessment.
Toward Foundation Models for Online Complex Event Detection in CPS-IoT: A Case Study
Technical report on using foundation models for online complex event detection in CPS-IoT environments.
NARCE: A Mamba-Based Neural Algorithmic Reasoner Framework for Online Complex Event Detection
Technical report on a Mamba-based neural algorithmic reasoning framework for online complex event detection.
Communication and Sensing: Wireless PHY-Layer Threats to Security and Privacy for IoT Systems and Possible Countermeasures
Survey-oriented article on security and privacy threats introduced by physical-layer sensing in IoT systems, with countermeasures and architectural trade-offs.
Towards Cyber Resilience against APTs
Technical report on cyber resilience against APTs, centred on intelligence, causal analysis, and adaptive defence mechanisms.
Preliminary Insights Into Resource-Constrained Neuro-Symbolic Causal Complex Event Processing
Paper on causal, neuro-symbolic complex event processing under resource constraints, relevant to edge systems and distributed operational environments.
Early Insights into Argumentation-Guided Causal Evaluation with the Help of LLMs
Early contribution on combining LLMs and argumentation to produce more structured and explainable causal evaluation.
Risk-aware classification via uncertainty quantification
Paper on risk-aware classification in which uncertainty quantification becomes part of the decision process rather than a secondary score.
I sistemi d'arma autonomi nel diritto pubblico
Matteo Frau book on AI, military autonomy, and universal legal limits, included here as a governance-oriented contribution on autonomous systems, security, and accountability.
Wherever I May Roam: Stealthy Interception and Injection Attacks Through Roaming Agreements
Study of how roaming partnerships can be abused to intercept traffic and inject messages more stealthily than many users and operators assume.
X-squatter: AI Multilingual Generation of Cross-Language Sound-squatting
Peer-reviewed extension of the sound-squatting line, with measurements over TLS certificates and PyPI packages to show how much of the phenomenon already exists on the live Internet.
ISP5G+
Project on innovative security paradigms for beyond-5G networks, spanning physical-layer protection, infrastructure security, and monitoring for heterogeneous mobile networks.
SCARPHASE
Privacy-enhanced security framework for cooperative driving, focused on cross-technology protection of availability, trustworthiness, and sensitive data.
EMBRACE
Project on energy-efficient communication methods for challenging environments, with a local UniBS contribution on PHY/MAC protection and privacy in adversarial networks.
Learning Robust Reward Machines from Noisy Labels
Paper on learning robust reward machines from noisy labels, relevant to agents operating under imperfect or adversarial supervision.
Neuro-Symbolic Fusion of Wi-Fi Sensing Data for Passive Radar with Inter-Modal Knowledge Transfer
Paper on neuro-symbolic fusion of Wi-Fi sensing data for passive radar, with transfer of knowledge across sensing modalities.
TeamCollab: A Framework for Collaborative Perception-Cognition-Communication-Action
Framework for modelling collaborative teams across perception, cognition, communication, and action in distributed and mission-critical settings.
Physical-Layer Privacy via Randomized Beamforming Against Adversarial Wi-Fi Sensing: Analysis, Implementation, and Evaluation
Comprehensive study of physical-layer privacy against adversarial Wi-Fi sensing, including analysis, implementation, and experimental evaluation.
Standard Contractual Clauses for Transfers of Personal Data to Third Countries
Book chapter by Giorgio Pedrazzi on the legal instruments that make international data transfers possible with GDPR-grade safeguards, relevant to cloud, incident response, and cross-border cyber governance.
NEACD
Project led by Federico Cerutti on neurosymbolic active cyber defence, aimed at extracting actionable intelligence from logs, TECHINT, and heterogeneous evidence streams.
CSI-Future
Project on joint communication and sensing through Channel State Information, with emphasis on interpretability, operational protocols, and privacy/security safeguards.
Assessing the Robustness of Intelligence-Driven Reinforcement Learning
Technical report on the robustness of intelligence-driven reinforcement learning approaches under dynamic and noisy conditions.
Sound-skwatter (Did You Mean: Sound-squatter?) AI-powered Generator for Phishing Prevention
Technical report on a multilingual AI system that generates sound-squatting candidates to proactively defend against phishing on domains and software packages.
ACRE
Project on AI, causality, and reasoning to improve the resilience of deceptive cyber-defence assets against advanced persistent threats.
Lost in Translation: AI-based Generator of Cross-Language Sound-squatting
Workshop paper on cross-language sound-squatting, using AI to expose phishing risks that move across languages.
DeepProbCEP: A neuro-symbolic approach for complex event processing in adversarial settings
Journal extension of the neuro-symbolic complex-event-processing line, explicitly aimed at adversarial environments.
Human-machine collaboration in intelligence analysis: An expert evaluation
Expert evaluation of human-machine systems for intelligence analysis, useful for designing effective interaction and calibrated trust.
HAI-PCCA
Project on effective collaboration between humans and AI agents, with learning across perception, cognition, communication, and action loops.
NS-CEP
Project on uncertainty-aware neuro-symbolic AI architectures for signal intelligence, insight generation, and foresight on complex events in contested environments.
Integrating CSI Sensing in Wireless Networks: Challenges to Privacy and Countermeasures
Paper on integrating CSI sensing into wireless networks without opening new avenues for unauthorised localisation and inference, with focus on privacy and countermeasures.
AI-based Sound-Squatting Attack Made Possible
Workshop paper showing how AI can make sound-squatting attacks practical, anticipating the later multilingual phishing line.
Handling epistemic and aleatory uncertainties in probabilistic circuits
Paper on representing and separating different forms of uncertainty in compact probabilistic models, with implications for reliability and risk evaluation.
Using DeepProbLog to perform Complex Event Processing on an Audio Stream
Technical report on a neuro-symbolic architecture that uses audio, neural networks, and probabilistic rules to detect complex events even with noisy data.
DI-P2SL
Project on protecting location privacy against unauthorised localisation based on 802.11 signal analysis and CSI.
Supporting Trustworthy Artificial Intelligence via Bayesian Argumentation
Methodological contribution on using Bayesian argumentation to support AI systems that are more transparent, auditable, and suitable for trust-sensitive settings.
Self-Aware Effective Identification and Response to Viral Cyber Threats
Interdisciplinary paper on identifying and responding to viral cyber threats, with emphasis on systems that understand their own state and react adaptively.
DeepSQA: Understanding Sensor Data via Question Answering
Paper on applying question answering to sensor data, enabling more intelligible querying of complex streams in IoT and CPS settings.
Augmenting phishing squatting detection with GANs
Preliminary study on using GANs to expand the detection of domains and resources associated with phishing and squatting campaigns.
Frankenstein: Advanced Wireless Fuzzing to Exploit New Bluetooth Escalation Targets
Work on a wireless fuzzing framework that revives closed-source firmware in emulation to uncover real vulnerabilities in Bluetooth stacks and their interaction with host systems.
Rapid Trust Calibration through Interpretable and Uncertainty-Aware AI
A contribution on making AI systems more readable and more honest about their limits, so they can support cyber decisions without inviting blind trust.
Uncertainty-Aware Deep Classifiers Using Generative Models
A method for training deep classifiers that better separate known cases, ambiguous cases, and out-of-distribution inputs, reducing overconfidence in unsafe predictions.
No results match the current filters.