ML4A / Body Feedback

ML4A / Body Feedback

Body Feedback

This project has been developed during the one-week workshop "Machine Learning for artists" held by Gene Kogan in Opendot on November 2016.

/ Concept 
The goal is to design and develop a tool for physiotherapeutic treatment, to be used to stimulate motor activity in kids with reduced mobility due to neurological diseases. The main idea is to stimulate motor activity through sound feedback. The use of machine learning (neural networks) is necessary to adapt the interaction to the specific and peculiar motor abilities of the kids. The sonic system relies on a three dimensional map where different sounds are localised basing on their correlations using tSNE (https://github.com/opendot/ml4a-soundcube).

/ Realisation
We trained two main regression models, one intended for kids with reduced mobility
Project_SpaceDependent_setup.wekproj and one for kids able to do larger movements

The inputs and outputs used in each model are summarised in the table below. The BPM input is used as feedback of the experience and is shared among the two models. This parameter affects the magnitude (weight) of the movement in the virtual space.

/ Team
Vittorio Cuculo, Jordi Garreta, Emanuele Lomello, Gene Kogan, Alessandro Masserdotti, Andrea Rossi and Opendot