Led by particle physist and choreographer Dr. Mariel Pettee, our team of dance artists, physicists, and machine learning researchers has collectively developed several original, configurable machine-learning tools to generate novel sequences of choreography as well as tunable variations on input choreographic sequences. We use recurrent neural network and autoencoder architectures from a training dataset of movements captured as 53 three-dimensional points at each timestep. 

 

Our team’s vision for the future of creative artificial intelligence necessitates the modeling of not only written, visual, and musical thought, but also kinesthetic comprehension. In light of the modern understanding of movement as an intellectual discipline, the application of machine learning to movement research serves not as a mere outsourcing of physical creative expressiveness to machines, but rather as a tool to spark introspection and exploration of our embodied knowledge. The published paper can be found here.