2022 Diké workshop - Bias in (compressed) language Models

Program

Overview

This workshop will explore several notions of bias in Machine Learning, more precisely in Language Models, with applications including fairness, unbalanced learning and multilingual models. This day will be an opportunity to listen to high quality invited speakers that work in the domain and to discover the Diké collaborative project most recent findings. The presentations will be followed by a convivial time around French viennoiseries to open discussions and share ideas on the topic. This workshop is free, but registration is mandatory :

• Registration

Invited Speakers

Ian Davidson is a Professor in the computer science department of University of California, Davis. He works on AI, machine learning and data mining algorithm development. With domain expert collaborators Ian Davidson works on high societal impact domains such as neuroscience, intelligent tutoring systems and social network. His research has been funded by a variety of sources, such as grants from Google, Yahoo!, National science foundation, Department of Defense, and the Henry M. Jackson Foundation. He has received the NSF CAREER award and is on the editorial board of several transactional journals published in his area, such as IEEE and ACM. He recently gave a tutorial on Fairness in Machine Learning at KDD 21, and published several papers in the domain.

Jean-Philippe Magué is assistant professor in linguistics at the ENS de Lyon. He is a member of the Interactions-Corpus-Apprentissages-Representations lab and define himself as a Digital Humanist. The projects he is involed in lie in three categories: Computational sociolinguistics (he is PI of the SoSweet project), Scholarly text editing and digital publication (he was the leader of the Atelier des Humanités Numériques until 2015/12/31), and Digital heritage.

Aurélie Névéol is a CNRS Researcher at LISN (formerly, LIMSI) working on clinical and biomedical Natural Language Processing. Her research interests include information extraction and knowledge representation in specialized domains. Her research addresses both methods and applications of biomedical text analysis, ranging from explorations of representation models and their cross-language or cross-domain adaptability, to the integration of representation frameworks to extract new medical knowledge from clinical text.