This experimental project investigated how people perceive behaviours of reinforcement learning through sound.
Reinforcement learning defines a formal framework for the interaction between a learning agent and an environment in terms of states, actions, and rewards. Our interest lied in interactive approaches to reinforcement learning, where the agent directly receives the reward signal from human feedback. Such an expressive workflow could allow humans to teach agents specific behaviours based on subjective preferences.
We were interested in studying how people may perceive agent behaviour through sound listening during feedback-based teaching. We thus led a controlled experiment where we asked participants to guide three types of agents using only feedback, in two sound synthesis environment. In all cases, participants successfully interacted with these agents to reach a sonic goal. Subjective evaluations showed that agents’ exploration behaviour, rather than their learning to reach a goal, was critical to how participants heard them as collaborative.
The project was developed with Frédéric Bevilacqua and Baptiste Caramiaux in collaboration with the ISMM group of IRCAM and the ex)situ group of LRI (INRIA), in the context of the Sorbonne Université Doctorate in Computer Science.
Paper at SMC (2018)