© 2009-2022 Strange Loop
In a year marked by physical distance, my only dance partner was a past version of myself. Using motion capture of my own movements as training data, I have worked with teams of collaborators to develop custom machine learning tools, including a Variational Autoencoder (VAE) and Graph Neural Network (GNN), to generate choreography that feels like my own. In this talk, I'll discuss not only these models and their generative capabilities (and pitfalls), but also the creative motivations and tensions that guided me while shaping this research trajectory. This is a story about engaging with AI as a creative collaborator to understand to what extent algorithms can reflect the most cherished parts of our identities.
Mariel Pettee recently defended her PhD in Physics at Yale University and is now a Chamberlain Postdoctoral Fellow at Lawrence Berkeley National Lab. Her research encompasses the development of custom machine learning techniques for high-energy particle physics, with a particular emphasis on creating generic techniques that have broad applicability across other areas of fundamental science and art. As a choreographer, director, and performer, she also uses theater and dance work to research audience activation, duration, power, self-documentation, authenticity, fear, and playfulness. Since 2017, she has led independent teams of researchers across academia, industry, and the arts using machine learning to generate choreography based on 3D motion capture of her own movements. Prior to her PhD, she earned her Bachelors in Physics & Mathematics from Harvard University and her Masters in Physics at the University of Cambridge (Trinity College) as a Harvard-Cambridge Scholar.