Electroencephalography Adaptive Classification and Decoding Techniques

• Published in: 2nd Smart Cities Symposium (SCS 2019) p. 43
• Main Authors: Sallout, M, Mattar, E
• Format: Conference Proceeding
• Language: English
• Published: Stevenage, UK IET 2019; The Institution of Engineering & Technology
• Subjects: Bioelectric signals; Biology and medical computing; Classification; Classifiers; Digital signal processing; Disabilities; Electrical activity in neurophysiological processes; Electrodiagnostics and other electrical measurement techniques; Electroencephalography; Human-computer interface; Machine learning; Neural computing techniques; Signal processing and detection; Tensors; User interfaces; electroencephalography; human mind; adaptive classification approaches; transfer learning; dynamic classifiers; brain-computer interfaces; EEG-BCI; adaptive classification methods; physical disabilities; signal classification; medical signal processing; adaptive classifiers; deep learning mechanisms; decoding techniques; electroencephalography adaptive classification; brain computer interface; unsupervised adaptation; learning (artificial intelligence); matrices; tensor class
• ISBN: 9781839531071; 183953107X
• DOI: 10.1049/cp.2019.0216

Electroencephalography (EEG) classification is an essential component of Brain Computer Interface (BCI), which allows to communicate from the human mind to computer, and thus to communicate even for subjects with physical disabilities. There are various classes of classification methods related to EEG-BCI. Researchers have classified these types into four fundamental categories. The first is an adaptive classification approaches. The second is based on using matrices and tensor class of classification. The third is about the use of transfer learning, and final the fourth is about the use deep learning mechanisms. Given this background, this research framework provides a concise survey of adaptive classification methods employed for EEG based Brain Computer Interface. As indicated to, the adaptive classifiers, are dynamic classifiers where there parameters are incrementally re-evaluated and updated over time as new EEG data become available. In addition, the research frame has picked to establish an overall review to this specific category of classifier since, adaptive type of classifiers, have indicated to be superior to other static types of classifiers, as in reference to limited supervision or unsupervised adaptation, Lotte et. al. [1].