PhD in Modeling and Data Science

• Finding an equilibrium in Machine Learning Among Performance, Ethics (Privacy, Fairness), Explainability (Trust, Understandability) and Security (Robustness against Attacks and Confidentiality in Information)

(advisor: Rosa Meo, Department of Computer Science) 

• Stochastic Games in Continuous Time with Asymmetric Information and Learning

(advisor: Tiziano De Angelis, Department of ESOMAS)

Abstract: Stochastic games in continuous time are well-suited to describe strategic interactions across multiple agents. Examples of their applications arise naturally in economics, finance and operations research. In this project we will investigate how the access to different sources of information impacts the structure of equilibria. The study will be conducted using methods and techniques from stochastic calculus and stochastic control, combined with Bayesian learning.

• Stochastic modeling for random phenomena near critical thresholds with applications (possible
  co-tutele with University of Granada, Spagna)

(advisors: Elvira Di Nardo, Department of Mathematics, and Francisco de Asis Torres Ruiz, Univ. of Granada)

Abstract: The project focuses on the mathematical and computational analysis of systems characterized by stochastic behavior, particularly as they approach critical thresholds where their dynamics may undergo significant qualitative changes—such as transitions from stability to instability, from low to high volatility, or between different phases. These thresholds often represent change points that are crucial to understanding the evolution of complex systems. Although one-dimensional cases have been widely studied, emerging modeling needs requiring the extension of existing results in several directions, each motivated by both theoretical challenges and practical applications. The research will place particular emphasis on the study of first-passage times, the modeling of time-related variables that indicate critical transitions (e.g., pandemic peaks, energy resource depletion), and the use of higher-order cumulants and moment-based techniques to better describe system behavior near these thresholds. Additionally, the project aims to develop and implement efficient numerical methods and simulation algorithms to approximate the distributions of key random quantities.

• Dialogue Systems, Conversational Interfaces and Natural Language Generation for Artificial
  Intelligence

(advisor: Luca Anselma and Alessandro Mazzei, Department of Computer Science) 

Abstract: Natural language is the most sophisticated technique that machines can use in order to communicate with humans. Three fields of artificial intelligence that can highly benefit from Dialogue Systems (DS), Conversational Interfaces (CI) and Natural Language Generation (NLG) are (1) Automatic Reasoning, (2) Social Robot Interaction, (3) Assistive Technologies. Therefore, the project will focus on the study, the design and the application of DS, CI and NLG to these fields by integrating together both traditional symbolic/statistical models and modern large language models.

• Activity-driven modeling of financial transactions networks

(advisor: Marco Maggiora and Marco De Stefanis, Department of Physics, CENTAI)

• A Multi-Agent System approach for designing permeable Cities for pollinators

(advisor: Marco Maggiora and Marco De Stefanis, Department of Physics, CENTAI)

• Programming Models for Quantum Computing: Bridging Classical Patterns and Quantum Paradigms

(advisor: Marco Aldinucci, Department of Computer Science)

Abstract:  The project conceives quantum computing as an integral part of high-performance computing (HPC) systems. We seek project proposals that explore both quantum computing paradigms — quantum annealing-based and gate-based — to develop methodologies that, for example, adapt classical algorithmic patterns to describe the composition of quantum algorithms, or investigate the relationships between classical parallel computing and aspects tied to the non-determinism of quantum computational models. From an experimental perspective, we aim to evaluate the effectiveness of modular design on quantum simulators and native quantum systems, analyzing the potential of these technologies in the HPC context.

• Federated Learning as a Service: a coordination platform

(advisor: Robert Birke, Department of Computer Science)

• Bayesian nonparametric modelling of dynamic clustering

(advisors: Matteo Ruggiero and Giovanni Rebaudo, Department of ESOMAS)

• Kinetic interacting particle system of Stochastic Differential Equations and propagation of chaos

(advisor: Elena Issoglio, Department of Mathematics) 

ABSTRACT: This project focusses on the study of interacting particles whose position-velocity dynamics is random. More precisely, the evolution in time of the velocity of each particle depends on the velocity of all other particles and is perturbed by some noise, while the position of the particle is simply the integral of the velocity. This leads to a degenerate noise leading to models known as kinetic SDEs. We are interested to study well-posedness of these systems and their so-called propagation of chaos, namely the study of the limiting dynamics as the number of particles tends to infinity. 

• Algorithmic approach related to the exit time problem with modeling applications

(advisors: Cristina Zucca, Department of Mathematics and Samuel Herrmann (Université Bourgogne Europe)

• Soil moisture modelling for drought and flood forecasting, using ground and remote sensing data

(advisor: Stefano Ferraris, Interuniversity Department of Regional and Urban Studies and Planning)

Abstract: The project regards the application of statistical data analysis and stochastic models for forecasting hydrological events. They can be related either to excess of rainfall and snowfall, namely flooding, avalanches and landslides, or to lack of water, such as meteorological and agricultural droughts. Nowadays plenty of data are available, but they are quite recent and there is a lack of best practices for using them. Also, some of them have some important limitations, e.g. satellite soil moisture is available only to about 5 cm depth, and therefore it is necessary for most applications to infer the values at the depth of plant roots.

• Mathematical modeling of Anomalous diffusions.

(advisor: Bruno Toaldo,  Department of Mathematics) 

Abstract: The PhD student will have the opportunity to choose among various research directions within
this context: 1) Models associated with nonlocal PDEs (fractional equations, non-Markovian processes); 2) Diffusions with stochastic reflection; 3) Generative models (denoising diffusion) based on diffusion processes; 4) Simulation and Monte Carlo methods for anomalous diffusion processes