seminars

The potential of deep learning in genomic studies

Date
Tuesday, December 12, 2017

Time
11:00 a.m.

Location
ISI 2nd floor meeting room

Speaker(s)
Riccardo Scopigno

This is the era of “big data” and “big data” require new methods able to extract information from an ocean of useless data, that is, to transform  data into valuable knowledge. One crucial point in this process is the identification of the variables mostly influencing a given  phenomenon. For this reason, several new approaches are currently under investigation and are emerging in literature. One sector where the classic statistical approaches are not yet significantly sustained by alternative technologies is that of genomics.
While classic approaches have largely demonstrated their usefulness, by leading to all the theoretical genomic results available today, it is important to understand what possible relations could be currently neglected. For this reason, it is worth reflecting on the possible role of alternative approaches.
 The focus of the speech will be on machine learning instruments (and, in particular, on deep learning),  delving into their possible  effectiveness  in selecting features (variables)  and modelling the outcome, but also trying to identify the theoretical and practical gaps to be filled and paying attention to the harmonization with the classic and well-proven tools.


Biography
======
Riccardo Scopigno received the M.Sc. degree in Telecommunications Engineering (summa cum laude) with a thesis on Calculus (on variational problems with free discontinuities) and the Ph.D. degree in Information and Communication Technologies (further extending the topic of his M.Sc. thesis): both the degrees are from the University of Pavia (Italy), in 1995 and 2005 respectively. In 2017 he concluded a 2-year master course on “Statistics for Genomics” at the same University.

He owns a 20-year experience in ICT: he worked in Italtel-Siemens, In Marconi-Ericsson and has worked at ISMB as research leader since 2003. His current position is “Head of Research Area”, leading the MLW Department which he built from scratch and which deals with Wireless and Multimedia data processing.

He has coordinated and participated in many regional, national and European (FP7, H2020) projects set in manifolds application contexts (Industry 4.0, Smart cities, IoT, ITS, etc.), with a recent focus on transportation (road and rails).

His skills mainly cover telecommunications, multimedia and, recently, deep learning.

He is an IEEE Sr. Member and published about 100 papers, mostly on the application of new algorithmic solutions to  heterogeneous fields.