G. Roffo, S. Melzi, Online Feature Selection for Visual Tracking, The British Machine Vision Conference, BMVC 2016
G. Roffo, S. Melzi, Personality in Computational Advertising: A Benchmark, International Workshop on Emotions and Personality in Personalized Systems at ACM RecSys, EMPIRE 2016
G. Roffo, S. Melzi, Features Selection via Eigenvector Centrality, ECML/PKDD – New Frontiers in Mining Complex Patterns, NFMCP 2016
G. Roffo, S. Melzi, The Visual Object Tracking VOT2016 challenge results, IEEE European Conference on Computer Vision Workshops, ECCV 2016
S. Obertino, G. Roffo and G. Menegaz Infinite Feature Selection on SHORE based biomarkers reveals connectivity modulation after stroke, International Workshop on Pattern Recognition in Neuroimaging, PRNI 2016
G. Roffo, C. Giorgetta, R. Ferrario, W. Riviera and M. Cristani, Statistical Analysis of Personality and Identity in Chats Using a Keylogging Platform, ACM International Conference on Multimodal Interaction, ICMI 2014.
G. Roffo, C. Giorgetta, R. Ferrario and M. Cristani Just The Way You Chat: Linking Personality, Style And Recognizability In Chats , International Workshop on Human Behaviour Understanding, Associated with ECCV 2014.
G. Roffo, M. Cristani, L. Bazzani, H. Q. Minh, and V. Murino, Trusting Skype: Learning the Way People Chat for Fast User Recognition and Verification, IEEE International Conference on Computer Vision Workshops, ICCV 2013.
G. Roffo, C. Segalin, V. Murino and M. Cristani, Reading Between the Turns: Statistical Modeling for Identity Recognition and Verification in Chats, IEEE International Conference on Advanced Video and Signal- Based Surveillance, AVSS 2013.
G. Roffo, M. Cristani, F. Pollick, C. Segalin and V. Murino Statistical Analysis of Visual Attentional Patterns for Videosurveillance, Iberoamerican Congress on Pattern Recognition, CIARP 2013.
M. Cristani, G. Roffo, C. Segalin, L. Bazzani, A. Vinciarelli, and V. Murino, Conversationally-inspired stylometric features for authorship attribution in instant messaging, ACM international conference on Multimedia, ACMM 2012.
My primary research interests are in the area of pattern recognition and machine learning. Within these areas, my work focuses on developing novel strategies to formalize, explain and visualize the pattern in data. My work encompasses different case studies. Firstly, I developed powerful means for analysis and interpretation of sequential data coming from dyadic text chat conversations and for the extraction of interesting knowledge that could help in modeling social interactions and perhaps understanding the unique turn-taking dynamics such a system provides. Secondly, I worked on many aspects of feature selection in machine learning introducing the concept of graph-based feature selection. Such a framework is prone to parallelization making this family of algorithms highly scalable (i.e., suitable for Big Data analysis). Another important aspect I dealt with is the real-time factor of such techniques. This framework turns out to be also suitable for visual object tracking, demonstrating to improve tracking performance while maintaining high frame rates. Finally, I have also worked on soft-biometrics by taking into account user-centric aspects, such as personality. In fact, I exploited psychologically motivated constructs to infer personality traits from users writing behavior or, recently, to study the role of these cues in various aspects of recommender systems.
Online Feature Selection for Visual TrackingG. Roffo, S. Melzi, British Machine Vision Conference (BMVC 2016)
Personality in Computational Advertising: A BenchmarkG. Roffo, A. Vinciarelli, ACM RecSys - Emotions and Personality in Personalized Systems, (EMPIRE 2016) Oral Presentation
Features Selection via Eigenvector CentralityG. Roffo, S. Melzi, ECML/PKDD - New Frontiers in Mining Complex Patterns, (NFMCP 2016) Oral Presentation
The Visual Object Tracking VOT2016 challenge resultsG. Roffo, S. Melzi, Dynamic Feature Selection Tracking System, W21 at IEEE European Conference on Computer Vision, (ECCV 2016)
Infinite Feature Selection on SHORE based biomarkers reveals connectivity modulation after strokeS. Obertino, G. Roffo, G. Menegaz, IW on Pattern Recognition in Neuroimaging, (PRNI 2016) Oral Presentation
Infinite Feature SelectionIEEE International Conference on Computer Vision, 2015.
Reading Between the Turns: Statistical Modeling for Identity Recognition and Verification in Chats IEEE International Conference on Advanced Video and Signal-Based Surveillance, 2013. Oral Presentation
Conversationally-inspired stylometric features for authorship attribution in instant messagingACM International Conference on Multimedia, 2012.
Toolbox/Code & Datasets
TxT-Chat Database: A collection of more than 200 text chat conversations, collected by a chat service with key logging capabilities; this in practice allowed us to retain the timing of each single hit of a key (keystrokes + timestamp), recording at the finest level the behaviour of users while they type. In order to download the TxT-Chat database, please return a signed copy of the "Agreement" document to @Request.
Feature Selection Library: Feature Selection Library (FSLib v3.0 Last update August 2016) is a widely applicable MATLAB library for feature selection (attribute or variable selection), capable of reducing problem dimensionality to maximize the accuracy of data models, performance of automatic decision rules as well as to reduce data acquisition cost.
Research & Work Experience
Ph.D. Student - Intern
School of Computing Science, Glasgow (United Kingdom), Advisor: Senior Lecturer Alessandro Vinciarelli . I visited the School of Computing Science at the University of Glasgow, where I started a highly innovative project aimed at the automatic analysis of on-line text chats. In particular, I am developing computational approaches capable to predict personality traits and conflict handling style of subjects that negotiate using text chat platforms.
From Fundamentals to Appli- cations, Ragusa (Italy).
- As a participant I had the possibility to present the results of my research (poster n.62), and to interact with my scientific peers, in a friendly and constructive environment. I also benefited from direct interaction and discussions with world leaders in Computer Vision. http://svg.dmi.unict.it/icvss2014/
- Examination session successfully completed.
Pattern Recognition, and Image Processing, Genoa (Italy).
3R’s of Computer Vision: Recognition, Registration, Reconstruction, Ragusa (Italy).
G12 8QB Glasgow (United Kingdom), Internship.
Advisor: Prof. Frank Pollick, Co-Advisor: Prof. Alessandro Vinciarelli.
- A statistical analisys of visual attentional patterns for video-surveillance applica-tions, CIARP 2013.
Natural Languange Processing, Rome.
- Developed an automated information retrieval systems by using Apache Lucene in a 6 months collaboration.
- The search engine has been enhanced with machine learning techniques such as: LDA and PLSA.
Competition, Advisor: Prof. Andrea Fusiello.
- Shadow removal algorithm for textures by pyramid-based restoration process.
Department of Youth and the Italian Civil Service, Verona.
- Data analysis: from september 2005 to August 2006 I worked on the Enterprise Resource Planning (ERP) of the City Council of Verona.
Education and Training
Specific field in Visual Computing, University of Verona.
- Master’s programs: Machine Learning & Pattern Recognition, Computer Vision, Image-based 3D Reconstruction, among others.
- Thesis: A framework for automated video rating prediction using hybrid generative-discriminative classifiers
- Degree awarded in the first available session of the academic year.
Specific field in Multimedia,
University of Verona.
- Bachelor programs: Digital Image and Sound Processing, Computer Graphics, Object-oriented programming (OOP), Software Engineering.
- Thesis: Design and implementation of an interactive visualization system for video surveillance
- Degree awarded in the first available session of the academic year.