I am currently a researcher at Fondazione Bruno Kessler (Trento, Italy) in the group SpeechTek, investigating Deep Learning methods for Speaker Diarization and, in general, Acoustic Models.
I got my Ph.D. in Machine Learning in May 2017 at the Department of Information Engineering and Mathematical Sciences (DIISM), University of Siena (Siena, Italy), working under the supervision of Prof. Marco Gori and Prof. Marco Maggini. Hence, I collaborated as a researcher with the UDOO Team at the Mind Lab, Department of Social, Political and Cognitive Sciences (DISPOC), University of Siena (Siena, Italy) under the supervision of Prof. Antonio Rizzo, investigating Machine Learning solutions to perform Novelty Detection on Audio and Video Surveillance applications.
During the Ph.D. my research focused on Regularization Methods for Machine Learning, in particular we proposed the application of temporal coherence constraints to On–Line Learning, embedding the optimization process in the frameworks of Variational Calculus and Dynamic Systems. We explored the application of these methods to Graphs and Manifold Regularization in Semi-Supervised Learning scenarios, with the final aims of the implementation in deep architectures facing real-videos understanding tasks.