I am conducting interdisciplinary research at the intersection of Embodied Interaction and Machine Learning in a creative context.
With my students and colleagues, we create machine learning-based systems for understanding rich inputs such as human movements, and which allow for end-user appropriation in a creative setting. The challenge of the creative context is that the end-goal, in a given 'task', can be unclear to the user and may not be initially 'allowed' by the system. The key idea is to allow the system to adapt to the user and the user to be guided by the system. This is critical in movement-based practices like music and dance. Applications of this research include the design of movement-based digital musical instruments, technology-mediated dance movement learning, interactive systems for motor rehabilitation. In this application areas, typical constraints on the machine learning models we developed are: low data ressources (for training), fast inference (for testing), or interpretability. We used Bayesian models and different form of learning such as iterative learning, active learning or reinforcement learning. Finally, we also use these systems to research more theoretical questions on human behaviour such as the multifaceted nature of movement variability and its role on motor skill acquisition.
I received a PhD in Computer Science from University Pierre et Marie Curie and IRCAM in 2012. From 2012 to 2015, I worked at Goldsmiths College (University of London) as a research associate. In 2015, I worked for the London-based start-up Mogees Ltd as senior research scientist before starting a Marie Curie Research fellowship between McGill University and IRCAM from 2016 - 2017.