This project studied how professional creatives make use of alternative interactive supervised learning methods for prototyping motion-sound mappings.
Most interactive approaches to machine learning require humans to demonstrate examples to the machine to supervise its learning. However, examples might not be the best input modality in creative contexts, where users may not have clear specifications at the beginning of their design process.
We led a workshop with professional creatives using four alternative methods implemented in our « grab-and-play » software to build motion-sound mappings. The methods were shown complementary to example-driven interactive machine learning and shed a light on the importance of efficiency, exploration, and unpredictability in creative processes.
The project was developed with Rebecca Fiebrink in collaboration with the Department of Computing of Goldsmiths University of London, in the context of the ENS Paris-Saclay Pre-doctoral Research program.
Pre-doctoral Report (2016)