About me

I am a PhD student in Computer Science at the University of Trento. I am a member of the SML research group and my PhD advisor is Prof. Andrea Passerini.

I am currently working at Spotify London as a research intern, mentored by Mounia Lalmas and Rishabh Mehrotra.

Research

I work on several topics in Machine Learning, especially those concerning the prediction of structures and complex objects, as well as applications thereof. This research area is called structured-output prediction and is my current research focus. I’m interested in both the “old fashioned” large-margin approaches as well as recent deep architectures. I’m particularly interested in embedding knowledge into structured prediction models, usually through some kind of constraint solving/learning methods. During my PhD I applied this concept to preference-based decision problems over combinatorial spaces of choices. The resulting framework is called “constructive preference elicitation”, and will be the topic of my PhD thesis. My work touches several other areas of machine learning which I have also come to know pretty well, such as Online Learning, Convex Optimization, Preference Elicitation and Recommendation Systems. Other areas I am interested in are Deep Learning, Reinforcement Learning, Natural Language Processing, Dialogue Systems, Statistical Relational Learning and Constraint Programming. Check out my CV for more details.

Short Bio

I got my M.Sc. in Computer Engineering in 2015 from Sapienza University of Rome (Italy). During my last year of masters I studied for a semester at the University of Melbourne (Australia) and I worked on my masters thesis abroad at the University of Oslo (Norway). There I worked with Pierre Lison on semantics of dialogue and dialogue systems. After my graduation, I started my PhD at the University of Trento, a lovely small city in northern Italy, full of brilliant scientists. Here I started working with Andrea Passerini and Stefano Teso on constructive preference elicitation. During my PhD I also traveled a lot, spending some time in Paris, hosted by Paolo Viappiani at UPMC, and in Darmstadt, hosted by Kristian Kersting at TU Darmstadt. In June 2018 I started a research internship at Spotify Research London, where I work primarily on Contextual Bandits algorithms, together with Mounia Lalmas and Rishabh Mehrotra.

Contact

Paolo Dragone
Deptarment of Information Engineering and Computer Science - DISI
University of Trento
Via Sommarive, 9
38123 Povo (TN), Italy