作者: Alimed Celecia Ramos , Marley Vellasco
关键词:
摘要: In Brain-Computer Interfaces, one of the most relevant tasks is selection a subset features that efficiently describes EEG signal, excluding redundant and irrelevant features. This procedure reduces dimensionality dataset (avoiding curse) improves classification accuracy system. One successful models applied for this task use an Evolutionary Algorithm in wrapper approach. These produce excellent results but present drawback considerable high processing time, critical limitation its application on real Interfaces (BCI) systems. Quantum-inspired Algorithms can be alternative approach feature task, given they outperform classical exploration exploitation search space, obtaining global solution much faster. algorithm employs concepts principles from Quantum Mechanics to probabilistically describe set different states between logic 0 1. paper, developed tested over three subjects publicly available datasets. proposed model, Wavelet Packet Decomposition employed analyze time-frequency characteristics signals, Multilayer Perceptron Neural Network as classifier.