Description
Date depot: 20 mai 2022
Titre: Modèle d'adaptation dialogique par apprentissage par renforcement
Directrice de thèse:
Catherine PELACHAUD (ISIR (EDITE))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Intelligence artificielle
Resumé: This PhD focuses on human-agent interaction where the role of the agent is to optimize the amount of information to give to a user. In particular, the aim is to develop a dialog manager able to adapt the agent’s conversational strategies to the preferences of the user it is interacting with to maximize user’s engagement during the interaction. For this purpose, the idea is to propose a model where an agent is trained in interaction with a user using reinforcement learning approach. The engagement of the users will be measured using their non-verbal behaviors and turn-taking status. This measured engagement will be used in the reward function which balances the task of the agent (giving information) and its social goal (maintaining the user highly engaged). Different features will be considered such as the agent’s dialog acts that may have different impact on the user’s engagement depending on several factors, such as their personality, interest in the discussion topic, attitude toward the agent, and the type of nonverbal behaviors the agent will display. Objective and subjective studies will be conducted to evaluate the dialog manager model.
Doctorant.e: Galland Lucie