100 likes | 241 Views
Brain Computer Interfacing. Uses of the campus Grid in Cybernetics. Ian Daly, Dr Slawomir J. Nasuto , Prof. Kevin Warwick 17 th June 2009. What is a BCI. BCIs allow control of a computer by thought alone.
E N D
Brain Computer Interfacing Uses of the campus Grid in Cybernetics Ian Daly, Dr Slawomir J. Nasuto, Prof. Kevin Warwick 17th June 2009
What is a BCI • BCIs allow control of a computer by thought alone. • Allows individuals with severe motor impairments greater levels of communication and environmental control. • Uses: • Typing programs; Email, Text to speech, Twitter etc. • Environment; Lighting, TV, Wheelchair control etc. • Games; Table tennis, bio-feedback etc. • Prosthetics.
Types of BCI • Invasive vs. Non-invasive • Control vs. Goal orientated • P300 based • ERS / ERD based • Motor imagery
How it works • Stimuli presentation • Data recording • & pre-processing • Feature extraction • Training and classification http://www.musicandmeaning.net/issues/showArticle.php?artID=3.5 http://www.jvrb.org/archiv/760/index_html?set_language=en&cl=en http://ida.first.fraunhofer.de/projects/bci/competition_ii/albany_desc/albany_desc_ii.html
Our Research • Machine learning and signal processing • ICA, EMD, HMMs, Phase synchronisation • Artefact removal • Extraction of ERPs from single trials • Automated feature selection • Models for simulated ERP generation. • New types of BCI paradigm– speech imagery • Alternative hardware development
How we use Grid Computing (1) • Speech imagery • Template method investigated for classification of speech related EEG. • Large parameter space. • Multiple parameter subsets simultaneously evaluated on Condor. • Quickly able to demonstrate that template method over simplifies signal variability.
How we use Grid Computing (2) • Feature selection • EEG can be described by an infinite number of different features. • Feature selection algorithms - large search space. • GA’s • Swarm intelligence • Novel algorithms... • Condor allows quick traversal of the search space of possible features.
The Future • Need for newer / faster / more intuitive BCIs • Faster, more efficient control and communication • Greater ease of use • More robust and reliable • New BCI paradigms and more efficient algorithms in development. • Brain signal can be described in an infinite number of different ways. • Grid computing presents an effective way of investigating some of these possibilities.
Thank you for listening Questions? www.ucdmc.ucdavis.edu