A Physics-Informed GNN for Traffic Prediction: the DeepCTM Model
Published at ICANN 2026 in Padua (LNCS vol. 17091). A graph neural network that builds traffic physics into the model.
I'm Francesco, a PhD student in Methodological Statistics at Sapienza University of Rome, working on machine learning for graphs and bioinformatics.
I work where statistics meets machine learning: geometric deep learning on graphs, with applications to bioinformatics, traffic and environmental health.
I had the opportunity to study statistical methods in such different contexts, due to the double degree program between Sapienza and ENSAI. Starting from that experience I understood how to apply statistical techniques to real world problems.
Today my research moves between machine learning applied to biology, spatial and environmental health. We'll see what I can do!
Swimming and hiking in the mountains during free time, suffering through football matches :(
Published at ICANN 2026 in Padua (LNCS vol. 17091). A graph neural network that builds traffic physics into the model.
Presented at CIBB 2026. Extended version under review for the Springer LNBI post-proceedings.
A Bayesian graphical model used to study how the estimators converge.
Sapienza University of Rome, Italian National Institute of Health (ISS)
Rome. Machine learning for traffic flow and generative AI solutions on Azure.
Statistical Methods and Applications (110/110 with honors) and Smart Data Science. Thesis on traffic forecasting on the A90 highway.
Rome. Medical research projects in R and SPSS, and teaching applied statistics.
Sapienza University of Rome (108/110). Thesis on food insecurity during COVID-19.