Uni-Logo
You are here: Home BFNT Research A3: Co-Adaptivity of brain and machine: interaction of two learning systems
Document Actions

A3: Co-Adaptivity of brain and machine: interaction of two learning systems

Carsten MehringJ, Martin BogdanU, Bernhard SchölkopfV and Ad AertsenA


J = Animal Physiol. & Neurobiol, Inst. of Biol.I
A = Neurobiology and Biophysics
U = Wilhelm-Schickard-Institute for Computer Science, Tübingen
V = MPI for Biological Cybernetics, Tübingen
 


Scientific background

BMI approaches highly profit from adaptivity of the user’s brain and from adaptivity of the decoding algorithm. This approach also carries the danger that the two adaptive mechanisms might work against each other. The development of adaptive BMIs therefore will crucially depend on our understanding of how the two adaptive mechanisms of the human sensorimotor system and the machine interact with each other, and how this interaction can be controlled to achieve optimal performance. The interaction of the brain’s adaptivity with an adaptive decoder has only rarely been addressed and the computational mechanisms underlying such interaction are largely unknown.


Objectives

The goal of this project is to develop successful strategies and algorithms for co-adaptation in BMIs that can (i) improve BMI performance, (ii) increase learning speed, and (iii) increase the long-term stability of BMI control despite the non-stationarities of the neuronal activity. In addition, we aim to improve the understanding of human sensorimotor learning and control in adaptive dynamic environments.

Personal tools