Detection and explanation of individual and collective behaviour within a group to assess their wellbeing

• Title: ‘Detection and explanation of individual and collective behaviour within a group to assess their well-being’, Inria.
• PhD student: Sacha Germain.
• Affiliated unit: Inria, Matisse.
• Co-supervision:
  • Christine Largouët (INRIA IRISA Lacodam),
  • Charlotte Gaillard (INRAE Pegase),
  • Tassadit Bouadi (INRIA IRISA Lacodam),
  • Laurence Rozé (INRIA IRISA Lacodam)
• Doctoral school: Doctoral School of Mathematics, Telecommunications, Computer Science, Signals, Systems, Electronics
• Project duration: 2024 - 2027.

Project summary:

This thesis falls within the field of Artificial Intelligence, and more specifically machine learning on time series, with the aim of explaining individual behaviour. A herd of animals is composed of a group of individuals governed by different types of social relationships. The data available on these systems are mainly time series collected from numerous sensors. Their use makes it possible to capture behavioural dynamics, which are essential for estimating or learning many characteristics about individuals within their group. To assess well-being, it is necessary to monitor changes in the behaviour of individuals and the relationships they have with each other. This thesis therefore focuses on approaches that characterise the well-being of individuals by using data at the individual level, but also at the group level: (1) Propose methods for exploiting time series (such as discretisation, for example), (2) Develop an algorithm for learning the activity model at the individual level based on the timed automaton learning algorithm (activity trajectories associated with temporal constraints), (3) Integrate the concept of a group into a two-level representation, the individual level and the group level, using formalisms such as Relational Event Models (REM), and (4) Characterise the role and contribution of an individual in the collective dynamics of a group, drawing inspiration from game theory.