Biography
I spent the earliest part of my life in Rome. There, I completed my studies, obtaining a bachelor degree in Computer Engineering and a master degree in Artificial Intelligence & Robotics at Sapienza University of Rome. I graduated in 2012 with honours (cum laude).
In October 2012, I started my PhD in Machine Learning at Sapienza University of Rome.
In January 2014, I moved to Toronto, Canada, where I worked as Research Assistant at Ryerson University, while completing my PhD.
In February 2016, I began to work as a postdoc in Tromsø at UiT the Arctic University of Norway in the Machine Learning group.
In June 2017, I became a scientific collaborator at the Advanced Learning and Research Institute (ALaRI) at Universita' della Svizzera Italiana in Lugano, Switzerland.
In August 2018, I became a research scientist at NORCE Norwegian Research Centre, where I still work in a part-time position.
In September 2020, I became an Associate Professor at UiT, in the Dept. of Mathematics and Statistics.
Employment History
Tromsø, 2020 - Present
UiT The Arctic University of Norway
Associate professor at the Department of Mathematics and statistics.
Main research activities:
Pooling in Graph Neural Networks
Power flow balancing with Graph Neural Networks
Probabilistic forecasting with deep learning
Faults detection in power grids
Tromsø, 2018 - Present
NORCE - The Norwegian Research Institute
Senior Researcher. My work focuses on applying machine learning methods and developing new frameworks for the analysis of remote sensing and environmental data collected in the Arctic.
Main activities:
Detection of polar lows from SAR images
Detection of defects on solar panels in images taken from drones
Segmentation and classification of oil spills in SAR data with deep neural networks
Classification and segmentation of avalanches in SAR data with convolutional neural networks
Estimation of water level in lakes from airborne images
Tromsø, 2016 - 2018
UiT The Arctic University of Norway
Post-doctoral fellow at the Machine Learning Group at Physics and Technology department in University of Tromsø, Norway. My main research activity focused on Deep Learning methods, in the context of Recurrent Neural Networks and Autoencoders, Kernel methods, time series classification and data analysis on Electronic Health Records. My duties included the support of several PhD students in their projects.
Main activities:
Design of autoencoder trained with kernel methods for handling missing data.
Application of Recurrent Neural Networks on prediction tasks and development of new deep, recurrent architectures.
Development of novel deep learning approaches, based on kernel methods, for training Autoencoders.
Design of deep clustering algorithms, based on Information Theoretic Learning methods.
Kernel methods for clustering and classification of images and time series with missing data.
Application of Multiplex Horizontal Visibility graphs for the characterization of Recurrent Neural Network internal dynamics.
Lugano, 2017
Universita' della Svizzera Italiana
Scientific Collaborator at Advanced Learning and Research Institute (ALaRI) at the Faculty of Informatics in USI. My research focused on the analysis and classification of time series, relative to Health Data Records with Machine Learning methods.
Main Activities:
Employed in the research program "Hasler: A personal device for Automatic Evaluation of Health Status during Physical Training" for the development of machine learning algorithms for detection and prediction of paroxysmal atrial fibrillation events.
Toronto, 2014-2015
Ryerson University
Research Assistant at Ubiquitous and Pervasive Computing Lab. My research mainly focused on the development of a novel clustering-based procedure for knowledge discovery and on a particular class of recurrent neural network, called Echo State Network (ESN). During the two years, I have participated to a joint project on health care with the hospital in Toronto and I have supervised different master students during their thesis project.
Main activities:
Design and implementation of a multi-agent clustering system, conceived to identify and to exploit local dissimilarities among patterns.
Theoretical advances in reservoir computing (ESN) and application to time series forecasting of telephonic activities and electricity.
Theoretical studies on the criticality in complex, recurrent networks and development of unsupervised methodologies for effectively configuring the network.
Joint project with St. Michael Hospital in Toronto on automatic detection of faults in videos of laparoscopic surgery, by means of machine learning procedures.
Rome, 2012-2015
Sapienza University
PhD student in Machine Learning and Pattern Recognition at Department of Information Engineering, Electronics and Telecommunications. My PhD focused on the development of novel clustering techniques, applied in non-metric spaces, classification and graph matching. The PhD has been founded by Telecom Italia, with which I worked on a close collaboration on the analysis of Call Data Records, applying novel algorithms developed during my research activity. My PhD thesis is entitled Cluster analysis and applications on non-geometric spaces [link].
Main activities:
Design of a novel graph matching procedure, based on granular computing and evolutionary computation, with application to several classification problems, including images classification.
Design of a novel density-based clustering algorithm, operating in non-metric domains (sequences, graphs).
Design of a novel, highly parallelizable algorithm for the identification of subsequences and relative hardware implementation on embedded systems.
Applications in data analysis, data mining and knowledge discovery relative to the identification of relevant patterns in Call Data Records. My work has been done in a close collaboration with Telecom Italia and focused on: (1) profiling and characterization of users in a mobile network through a digital footprint; (2) design of functionalities in a Personal Data Storage system; (3) implementation of an anomaly detection system.