Veronica Guidetti

  skype:    veronica.guidetti@hotmail.it   

Biography

I’m a postdoctorate research fellow at the FIM Department of the University of Modena and Reggio Emilia. Within the FIM Department I collaborate closely with Professor F. Mandreoli, F. Motta and G. Buzzega. Our group is focused on Data Mining and Artificial Intelligence, and their applications in healthcare. I achieved a master (2016) and PhD degree (2021) in Theoretical Physics (FIS-02) at the University of Bologna. From October 2020 to November 2022, I was employed as a postdoc fellow at the Deutsches Elektronen-Synchrotron (DESY) in Hamburg, Germany. My detailed CV can be found at this link .

Interests and selected publications

My research at UniMoRe focuses on interpretable Machine Learning and its applications in high stakes domains, particularly in the medical field. Specifically, together with Professor Mandreoli (UniMoRe) and Davide Ferrari (King's college, London) we have created a multi-objective symbolic regression method capable of creating and managing non-linear, intelligible and parsimonious indexes in a data-driven way. Our goal is to make this and other interpretable techniques reliable, implement them in the everyday clinical practice and export them to federated learning contexts.

My main research interests in Deep Learning are the study of generalization and noise (see, e.g., double descent phenomenon, neural collapse, Effective theory of deep learning, information bottleneck). I am currently working on these topics with Nayara Fonseca (IBM Research, UK). Furthermore, I am interested in the development of optimizers based on physical models and in improving methods to solve differential equations using Physics-Informed Neural Networks PINNs. I am currently working on these topics with some members of IBM Research in UK.

My research activity in Physics focuses in particular on Axion-like particles, cosmological inflation and its effective field theories (EFTs) deriving from String Theory. In this regard, starting from this work, I am developing a Bayesian analysis method that allows to classify string inflationary models, understand which compactifications are experimentally favored and understand the impact of and the hierarchy between experimental and theoretical constraints. My main collaborators on these topics are Alexander Westphal (DESY, Hamburg) and Nicole Righi (King’s college, London).

Finally, I’m interested in ML applications to Finance. I’m working on data driven smart beta and alpha-factor definition to automatize portfolio management. My main collaborators on this topic are Dr. Alessandra Insana (University of Messina) and Dr. Francesco Muia (Oxford University).

This is a selected list of my recent publications on these topics:

• Guidetti, V., Righi, N., Venken, G., & Westphal, A. (2022). Axionic Festina Lente. Journal of High Energy Physics, 2023 (in publication)
• Davide Ferrari, Veronica Guidetti, Federica Mandreoli: Multi-Objective Symbolic Regression for Data-Driven Scoring System Management IEEE International Conference on Data Mining (ICDM), Orlando (FL), Dec. 2022. Open source code available on GitHub ;
• Davide Ferrari, Veronica Guidetti, Yanzhong Wang, Vasa Curcin: Multi-objective Symbolic Regression to Generate Data-driven, Non-fixed Structure and Intelligible Mortality Predictors using EHR: Binary Classification Methodology and Comparison with State-of-the-art AMIA Annual Symposium Proceedings. Vol. 2022. American Medical Informatics Association, 2022. Open source code available on GitHub;
• Mandreoli Federica, Ferrari Davide, Guidetti Veronica, Motta Federico, Missier Paolo (2022). Real-world data mining meets clinical practice: Research challenges and perspective. Frontiers in big Data, 99.
• Cicoli, M., Guidetti, V., Muia, F., Pedro, F. G., & Vacca, G. P. (2022). On the choice of entropy variables in multifield inflation. Classical and Quantum Gravity, 39, 195012.
• Cicoli, M., Guidetti, V., Muia, F., Pedro, F. G., & Vacca, G. P. (2022). A fake instability in string inflation. Classical and Quantum Gravity, 40, 025008.
• Cicoli, M., Guidetti, V., Righi, N., & Westphal, A. (2022). Fuzzy Dark Matter candidates from string theory. Journal of High Energy Physics, 2022(5), 1-52.

You can find the full list of my publications here , my Google Scholar profile here .