Umberto Junior Mele
PhD Candidate in Informatics, Specializing in Artificial Intelligence applied to Operation Research
Welcome to my personal corner on the web! I'm Umberto, a dedicated researcher and PhD candidate in Informatics at the Università della Svizzera Italiana, with a deep passion for artificial intelligence and its transformative potential.
I embarked on my academic voyage at the prestigious Sapienza Università di Roma, where I earned my Bachelor's in Management Engineering and my Master's in Data Science. My master's thesis, focusing on statistical inference for latent stochastic differential mixed effects models, laid the groundwork for my current PhD research. Now, at the Università della Svizzera Italiana, I am delving into the application of machine learning to combinatorial optimization under the guidance of renowned experts Prof. Roberto Montemanni and Prof. Luca Maria Gambardella.
Machine Learning in Combinatorial Optimization
At the heart of my research lies a fascinating convergence between artificial intelligence and combinatorial optimization. This interdisciplinary venture aims to revolutionize traditional optimization methods through the integration of advanced machine learning techniques. My journey in this domain is fueled by a quest to harness the power of AI to solve complex optimization problems, a challenge that has profound implications across various industries.
My current PhD research, undertaken at the Università della Svizzera Italiana, centers on applying machine learning to enhance combinatorial optimization solvers. The primary objective is to develop innovative frameworks that integrate predictive models and optimization algorithms, thereby increasing the efficiency and effectiveness of solving complex problems. This research not only pushes the boundaries of existing knowledge but also opens new avenues for practical applications in fields such as logistics, transportation, and resource management.
Key Contributions and Findings
My contributions can be summarized in the following three main areas:
Machine Learning-Driven Heuristics: My work includes developing new constructive heuristics driven by machine learning for the Traveling Salesman Problem (TSP), one of the most studied problems in combinatorial optimization. These heuristics leverage predictive models to make informed decisions during the solution construction phase, leading to more effective solutions.
Reinforcement Learning in Vehicle Routing: Another significant aspect of my research involves using reinforcement learning techniques to tackle vehicle routing problems.
Integrative Methods for Optimization Challenges: I am also exploring the integration of machine learning in local search and meta-heuristic strategies for various combinatorial optimization problems. These integrative methods aim to combine the strengths of AI and traditional optimization techniques, leading to more robust and scalable solutions.
As I continue my research journey, my goal is to further explore the potential of machine learning for reshaping the landscape of combinatorial optimization. The ultimate aim is to create systems that not only solve existing problems more efficiently but also tackle new challenges that were previously considered intractable. Through continuous innovation and interdisciplinary collaboration, I aspire to contribute to the evolution of AI as a pivotal tool in optimization and beyond.