Abstract
This work describes a generalized method for classifying motor-related neural signals for a brain-computer interface (BCI), based on a stochastic machine learning method. The method differs from the various feature extraction and selection techniques employed in many other BCI systems. The classifier does not use extensive a-priori information, resulting in reduced reliance on highly specific domain knowledge. Instead of pre-defining features, the time-domain signal is input to a population of multi-layer perceptrons (MLPs) in order to perform a stochastic search for the best structure. The results showed that the average performance of the new algorithm outperformed other published methods using the Berlin BCI IV (2008) competition dataset and was comparable to the best results in the Berlin BCI II (2002–3) competition dataset. The new method was also applied to electroencephalography (EEG) data recorded from five subjects undertaking a hand squeeze task and demonstrated high levels of accuracy with a mean classification accuracy of 78.9% after five-fold cross-validation. Our new approach has been shown to give accurate results across different motor tasks and signal types as well as between subjects.
Highlights
A brain-computer interface (BCI) is a system that measures central nervous system activity and converts it into an artificial output, such as movement of a cursor or robotic limb, thereby providing an alternative pathway for the brain to interact with its environment [1]
We have demonstrated the feasibility of a new method for creating subject-specific motor brain-computer interfaces (BCIs) that do not rely on a-priori selection of features or recording channels
The use of a high-dimension neural network has been investigated as a method for the classification of real-time EEG for use in a BCI
Summary
A brain-computer interface (BCI) is a system that measures central nervous system activity and converts it into an artificial output, such as movement of a cursor or robotic limb, thereby providing an alternative pathway for the brain to interact with its environment [1]. The signal that is used for BCIs is a measurement of the electrical or magnetic fields of the brain [2]. A grand challenge in BCI research and development, along with more general fields in neuroscience, is developing methods to decode a user’s intent from the neural signals. An ideal solution would be a method using a portable, non-invasive recording system, such as PLOS ONE | DOI:10.1371/journal.pone.0131328. A Generalizable BCI Using Machine Learning for Feature Discovery An ideal solution would be a method using a portable, non-invasive recording system, such as PLOS ONE | DOI:10.1371/journal.pone.0131328 June 26, 2015
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call