Interest in developing a new method of man-machine communication - the Brain-Computer Interface (BCI) - has grown steadily over the past few decades. A BCI is a communication system that implements the principle of ``think and make it happen without any p hysical effort'' by bypassing the conventional output pathways of nerves and muscles. BCI technology therefore can provide a new communication and control option for individuals that cannot otherwise express their wishes to the outside world. From a mach ine learning standpoint BCI presents a number of significant challenges. Firstly, the features produced by the brain monitoring system are multi-dimensional (10s sensors), rapidly sampled (100-1000s Hz), time series. Secondly, short training times imply that only a few examples (trials of the training session) are available. A third issue is the high level of noise in the signals from both non-neural and neural sources. A final challenge is the high variability of brain signals, due either to changes in the users non-BCI relevant brain states or users changing strategies to improve BCI control. I propose to investigate integrating adaptivity into the machine learning process. The objective here is to determine if integrating adaption with powerful clas sifiers is feasible and whether the potential benefits of adapting to the changing BCI environment are realisable. The project will be undertaken at a European centre of excellence for Machine learning and BCI, under the guidance of world leaders in the f ield. The training in machine learning and BCI that I will receive will be essential to a career in this emerging field.
Call for proposal
See other projects for this call