Academic Interests
My research interests lie at the intersection of uncertainty quantification theory with Conformal Prediction (CP) and graph machine learning, in particular how statistical coverage guarantees can be integrated into high-dimensional, structured domains like Knowledge Graphs (KGs). My work will explore the duality between Graph Neural Networks (GNNs) and symbolic logic, aiming to build models that are not only predictive but also logically consistent and trustworthy. Beyond these core areas, my interests are related to the latest trends of AI, including Agentic AI and multimodal LLMs.
PhD project
Logic-Aware Conformal Prediction for Knowledge Graphs.
This project aims to develop a novel framework for quantifying prediction uncertainty with data which must be compliant with a set of logical constraints by integrating Conformal Prediction with Graph Neural Networks. The research focuses on designing new algorithms that provide statistical coverage guarantees that exploit these constraints to give more reliable outputs for complex tasks such as link prediction and node classification. A key objective is to make these methods "logic-aware," leveraging the inherent connections between GNNs and symbolic reasoning to ensure that predictions are not only statistically sound but also logically consistent.
Background
I obtained both my bachelor and master degrees at 'La Sapienza', University of Rome, respectively in 2022 and in 2024. I gained some industry experience as software engineer in the Air Traffic Management sector.