| 2 |
* Dopo la pubblicazione del bando sono state formalizzate 4 borse finanziate dall'AGENZIA PER LA CYBERSICUREZZA NAZIONALE su specifiche tematiche: 1) “YOSO MPC at Scale” - Supervisore: Prof. Daniele Venturi - Abstract: Secure Multiparty Computation (MPC) allows mutually distrustful parties to jointly compute a function on their inputs privately, revealing only the output. This project aims to develop new MPC protocols for massively distributed contexts with dynamic participation, such as blockchains and machine learning. The emerging YOSO framework, where participants only engage briefly, allows lower-resource parties to contribute, democratizing private computation. Current YOSO protocols, however, have limitations, such as requiring a strong honest majority and being inefficient for specific tasks. The project seeks to design more efficient YOSO protocols for concrete functions, tolerating higher corruption thresholds.; 2) “Modelli di binary similarity per la rivelazioni di
vulnerabilità in firmware e binari resistenti ad attacchi avversari” - Supervisore: Prof. Giuseppe Antonio Di Luna - Abstract: The Internet of Things (IoT) revolution is introducing many smart devices into various sectors, but they often lack security due to firmware vulnerabilities, posing risks to users and society. This research project aims to use neural networks to identify vulnerabilities in real software, optimize their performance, and enhance their resilience to adversarial attacks. The project will focus on comparing current models, evaluating their resistance to attacks, and developing a robust binary similarity model over three years.; 3) "Attacchi e Sicurezza nei Foundation Models" - Supervisore: Prof. Giuseppe Francesco Italiano - Abstract: Foundation Models (FMs) are versatile machine learning models used in applications like image recognition, medical diagnosis, and generative AI. They leverage deep learning, are trained on vast datasets, and are complex, making them vulnerable to various security threats. This research aims to understand and mitigate these vulnerabilities, focusing on adversarial attacks, data poisoning, and prompt injection, to ensure the models' reliability and safety.; 4) “SPoND: Security and Privacy of Networks of Drones” - Supervisore: Prof. Riccardo Lazzeretti - Abstract: Unmanned Aircraft Vehicles (UAVs), or drones, are significantly impacting the European economy and society due to their versatility, technological advancements, and cost reductions. Initially developed for military purposes, drones have gained popularity in commercial and public sectors, and their use is expected to grow in areas like transport and services. The project aims to address security and privacy issues in drone networks with innovative solutions such as secure cooperation protocols, device identification and authorization, detection of compromised devices, and dynamic firmware analysis. |