ML_INFN is the national initiative of INFN to coordinate the effort towards the widespread application of machine learning technologies to the research activities on Particle Physics, Nuclear Physics and their applications in particular to medicine (Medical Physics).
The Florence group of ML_INFN is a node in vast collaboration network that connects research groups on similar physics challenges with world leading experts in computing and software technologies. We operate mainly in three fields:
- Data Analysis and Fast Simulation for Particle Physics. To employ machine learning technologies to data analysis a deep undestanding of the statistical meaning and reliability of input data and computed quantities is crucial. On the other hand, the simulation of the response of particle detectors is tremedously expensive in terms of computing resources and can be accelerated by using techniques and technologies widely applied for deep learning. For more on this, check for example the DeepHep-ino initiative. These activities are enhanced by a stric collaboration with CERN, the European Center for Particle Physics.
- Telemedicine and personalized medicine. Machine learning can back clinicians in the interpretation of clinical results. INFN operates since many years in strict collaboration with the Regional Health System to study and develop radiotherapy, nuclear medicine and radiation-related diagnostics. The Florence group is currently exploring machine learning technologies to provide the clinicians valuable hints in the interpretation of CTs and dose plans. The University of Florence and the Careggi Hospital support these initiatives.
- Quantum-inspired algorithms. Quantum physics provides a wide set of tools to describe optimization problems. This formalism can be used to design machine learning algorithms for standard computer, which are usually extremely fast and compressed with respect to deep neural networks, but also to design machine learning algorithm specific for quantum computing architectures. These activities are deeply connected with the Theoretical Physics Group of the Department of Physics and Astronomy of the University of Florence.
- Analysis of Environmental Pollution. Accelerating ions on a target and studying the expelled atoms is a powerful technique to study the elemental composition of the target. Exposing a filter to the air and then studying its elemental composition is an incredibly precise technique to identify the pollutants. The LABEC laboratory in Sesto Fiorentino is a world leader in the development of environment monitoring techniques based on nuclear physics and is a major contributor in the application of machine learning to automate the analysis of the measured spectra.
Contact person: Lucio.Anderlini [at] fi.infn.it
National web page: https://confluence.infn.it/display/MLINFN/Entry+Point+ML-INFN