BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2// METHOD:PUBLISH X-WR-CALNAME;VALUE=TEXT:Eventi DIAG BEGIN:VTIMEZONE TZID:Europe/Paris BEGIN:STANDARD DTSTART:20241027T030000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET END:STANDARD BEGIN:DAYLIGHT DTSTART:20240331T020000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT UID:calendar.28343.field_data.0@www.diag.uniroma1.it DTSTAMP:20260415T025127Z CREATED:20240612T095726Z DESCRIPTION:We are pleased to announce an upcoming event dedicated to promo ting women's participation in Data Science and related Ph.D. programs.Plea se find the program of the event below:15.00 Welcome address15.10 A Theore tical Analysis of Recommendation Loss Functions under Negative SamplingDr. Giulia Di TeodoroBio:Giulia Di Teodoro is a postdoctoral researcher at th e University of Pisa\, within the Information Engineering Department. She earned her degree in Management Engineering with honors from Sapienza Univ ersity of Rome\, Italy\, where she also completed her PhD in Data Science in 2024. Her doctoral thesis focused on precision medicine for HIV and dia betes as well as the interpretability of machine learning models. Giulia h as publications in the fields of Precision Medicine\, Explainable Artifici al Intelligence\, and Bioinformatics. Her research interests include Mixed -Integer Linear Programming (MILP)\, Precision Medicine\, and Recommendati on Systems.Abstract: Recommender Systems (RSs) are pivotal in diverse doma ins such as e-commerce\, music streaming\, and social media. This work con ducts a comparative analysis of prevalent loss functions in RSs: Binary Cr oss-Entropy (BCE)\, Categorical Cross-Entropy (CCE)\, and Bayesian Persona lized Ranking (BPR). Exploring the behaviour of these loss functions acros s varying negative sampling settings\, we reveal that BPR and CCE are equi valent when one negative sample is used. Additionally\, we demonstrate tha t all losses share a common global minimum. Evaluation of RSs mainly relie s on ranking metrics known as Normalized Discounted Cumulative Gain (NDCG) and Mean Reciprocal Rank (MRR). We produce bounds of the different losses for negative sampling settings to establish a probabilistic lower bound f or NDCG. We show that the BPR bound on NDCG is weaker than that of BCE\, c ontradicting the common assumption that BPR is superior to BCE in RSs trai ning. Experiments on three datasets and two models empirically support the se theoretical findings.15.50  Doing research in the industry and a path t owards academia.Dr. Aleksandra PiktusBio:Aleksandra is currently a second year PhD student in data science at Sapienza and a research engineer worki ng on large language models at Cohere. Before that she was a researcher at HuggingFace and the Facebook AI Research lab in London. She started her c areer as a software engineer at Facebook where she worked on reducing the spread of misinformation on the platform and on search. Her interests incl ude knowledge-intensive NLP\, retrieval augmentation and LLM pre-training data exploration.16.30  Engineering...? why?Prof. Marilena VenditelliBio:M arilena Vendittelli obtained her PhD in Systems Engineering in 1997 from t he University of Rome 'La Sapienza'. She won two Marie Curie Research Trai ning Grants in 1996 and 1998\, respectively\, for research on motion plann ing and control of nonholonomic systems during her postdoctoral period at LAAS-CNRS in Toulouse (France). From 1998 to 2016 she was at DIAG\, first as a post-doc student and then as Assistant Professor (2001-2016). From 20 17 to 2019 she was an associate professor at the Department of Information \, Electronics and Telecommunications Engineering and in 2020 she (re)join ed the DIAG. Over the years she has been a Visiting Scholar at Carnegie Me llon University (2005)\, the Courant Institute of New York University (201 2)\, the Simons Institute of UC Berkeley (2016).Abstract: The most common reaction a girl encounters when she says she wants to study engineering at university is one of surprise\, sometimes mixed with admiration\, followe d by the question\, 'Why?' In this talk\, I will humorously revisit the be ginnings of my engineering career and the non-technical challenges a girl faces in this field. Then\, I will briefly present some of my recent engin eering projects\, which are the result of my research in the field of robo tics. DTSTART;TZID=Europe/Paris:20240613T150000 DTEND;TZID=Europe/Paris:20240613T150000 LAST-MODIFIED:20240612T101847Z LOCATION:Aula Magna DIAG SUMMARY:Promoting Women's Participation in Data Science - Giulia Di Teodoro \, Aleksandra Piktus\, Marilena Vendittelli\n\n\n \n \n\n \n\n\nFabri zio\n\n\nSilvestri \n\n \n\n \n\n\n\n\n\nProfessore ordinario\n\n\npa gina personale\n\nstanza: \n\nB209\n\ntelefono: \n\n+39 0677274015\n\nMemb er of: \n\n \n\n \n\n \n\nBiografia: \n\n\n\nFabrizio Silvestri is a Full Professor and the coordinator of the Ph.D. in Data Science\, at Dipar timento di Ingegneria informatica\, automatica e gestionale (DIAG) of the University of Rome\, La Sapienza. His research interests lie in Artificial Intelligence\, and in particular\, Fabrizio Silvestri deals with machine learning applied to web search problems and natural language processing. H e is the author of more than 150 papers in international journals and conf erence proceedings. It holds nine industrial patents. He is the holder of the 'test-of-time' award at the ECIR 2018 conference for an article publis hed in 2007. He is the holder of three best paper awards and other interna tional awards. Fabrizio Silvestri spent eight years abroad in industrial r esearch laboratories (Yahoo! and Facebook). At Facebook AI\, Fabrizio Silv estri has directed research groups to develop artificial intelligence tech niques to combat malicious actors who use the Facebook platform for malici ous purposes (hate speech\, misinformation\, terrorism\, etc.) Fabrizio Si lvestri has a Ph.D. in computer science awarded by the University of Pisa with a thesis entitled: 'High-Performance Issues in Web Search Engines: Al gorithms and Techniques.'\n\n\nqualifica_rr: \n\nProfessors URL;TYPE=URI:https://www.diag.uniroma1.it/node/28343 END:VEVENT END:VCALENDAR