Riding the Co-Explorers
Résumé
We demonstrate Co-Explorers, a set of interactive agents that collaborate with users in musical space exploration. It implements an interactive reinforcement learning workflow, in which users can guide navigation by giving positive or negative feedback to the agent. It uses a deep reinforcement learning algorithm that optimizes agent exploration toward parts of space that might be of interest for users. Co-Explorers are independent of the type of space explored; the installation that we propose invites attendees to immerse themselves into collaborative sound space exploration.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...