BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Des approches d'apprentissage automatique pour la modélisation et
  la prédiction de la performance dans les sports collectifs / Machine Lea
 rning approaches to model and predict performance in Team Sports
DTSTART:20260709T073000Z
DTEND:20260709T110000Z
DTSTAMP:20260728T162500Z
UID:indico-event-16960@indico.math.cnrs.fr
DESCRIPTION:Speakers: Arnaud Odet\n\nJury composé de :\nLéo GERVILLE-REA
 CHE\, Rapporteur\, Université de BordeauxChristophe LEY\, Rapporteur\, Un
 iversité du LuxembourgBrigitte GELEIN\, Examinatrice\, ENSAI RennesBéatr
 ice LAURENT-BONNEAU\, Examinatrice\, INSA ToulouseSébastien DEJEAN\, Dire
 cteur de thèse\, Université de ToulouseCristian PASQUARETTA\, Co-directe
 ur de thèse\, Université de Toulouse\nIl sera également possible d'assi
 ster à la soutenance en visio :\nhttps://rendez-vous.renater.fr/Soutenanc
 e_Arnaud_Odet_623fa6-73d9ac-b3c9e0\nRésumé :\nAs sporting events and com
 petitions occupy an ever-increasing place in contemporary society\, with p
 rofessional athletes’ salaries\, media deals\, and related revenues reac
 hing unprecedented heights\, sport analytics research has blossomed accord
 ingly in recent years. As the stakes grow higher\, the demand for performa
 nce optimization in sports has never been more pressing. In parallel\, the
  latest advances in computer science\, combined with the ever-increasing c
 apacity of computing hardware\, have propelled Machine Learning into the s
 potlight as a potential new industrial and societal revolution. Deep gener
 ative models are interfering more and more with our daily lives\, and Arti
 ficial General Intelligence can hardly be dismissed as a mere fantasy anym
 ore.In this context\, this thesis aims to leverage recent advances in Mach
 ine Learning to provide practical and actionable tools to ultimately impro
 ve collective performance in team sports.At first\, we advocate that combi
 ning Machine Learning predictive power to the task of forecasting games ou
 tcomes and algorithms explainability techniques yields promising results. 
 Specifically\, we present a framework providing multiscale diagnostic anal
 yses that offer valuable insights regarding : (i) the drivers of a game ou
 tcome\, (ii) the identification of strengths and weaknesses of a given tea
 m\, and (iii) a general strategic understanding of the game. We illustrate
  this framework on men’s rugby union and women’s basketball. We then d
 emonstrate that incorporating players’ individual tracking data into dee
 plearning time series models increases our ability to forecast the short-t
 erm performance of their team. Applying feature permutation methods\, we h
 ighlight which players and features bear the greatest importance to this f
 orecasting task\, thereby characterizing their respective roles in play de
 velopment.Finally\, we adopt unsupervised machine learning to abstract pla
 yer archetypes from individual player statistics\, and leverage these arch
 etypes to identify the best-performing lineups (i.e. combinations of playe
 rs). Our results suggest that players’ complementarity prevails over jux
 taposition of individual talents. Adopting a broader angle and focusing on
  rosters (i.e. registered players for a team) instead of lineups\, we furt
 her emphasize thatstability is positively correlated with long-term perfor
 mance.Our work opens up several promising research directions\, including 
 the use of reinforcement learning to construct optimal rosters\, the ident
 ification\, and potential design\, of the most effective set plays and tac
 tical systems\, the modelling of tracking and event data through graph net
 works\, and the integration of emerging technologies. These perspectives a
 re extensively discussed in the dedicated section. We conclude that while 
 Machine Learning brings tremendous opportunities to improve performance in
  team sports\, interdisciplinary collaboration is required to fully unveil
  and achieve its potential.\n\nhttps://indico.math.cnrs.fr/event/16960/
LOCATION:Amphithéâtre Laurent Schwartz\, bâtiment 1R3 (Institut de Math
 ématiques de Toulouse)
URL:https://indico.math.cnrs.fr/event/16960/
END:VEVENT
END:VCALENDAR