Learning the Language of the Brain: Adaptive Brain-Machine Interfaces Maximize Information Transfer Through Autonomous Interaction with Brain Tissue
Freie Universität Berlin
Institut für Informatik
Dahlem Center for Machine Learning & Robotics
Gefördert von der Volkswagen Stiftung unter Az.: 97 545
Today's Brain Machine Interfaces (BMIs) are limited by our lack of knowledge about the internal dynamics of neurons and neural networks, yet a couple of successful clinical applications like electrical brain stimulation and neuroprosthetics have already emerged. The require manual customization and patient training, and are based on models that have to build in assiduous experimental work. We propose that a machine learning system could find meaningful stimulation parameters much quicker, by simultaneously building an internal model of the neural tissue, and maximizing the bandwidth of the channel into the biological networks.