Optimizing the number of electrodes and spatial filters for Brain-Computer Interfaces by means of an evolutionary multi-objective approach

摘要: Optimization of channel selection and spatial filter for Brain Computer Interfaces.A multi-objective evolutionary algorithm simultaneously optimizes channels.Multi-objective approach returns a set solutions (channels vs. error tradeoffs).A simple threshold is encoded to select channels instead longer binary mask.Joint optimization performs better than only the channels. Obtaining high accuracy classification from Interfaces require attach many electrodes on scalp subjects. On other hand, their placement involves generally laborious time consuming process. Therefore, it important practitioner estimate how electrodes, which ones, are needed obtain required accuracy. With this purpose, formulation proposed in order (Pareto front) that represent all optimal tradeoffs between number accuracy, where can choose. Additionally, previous research has shown highly depends proper tuning used preprocess electroencephalogram. work, Non-dominated Sorting Genetic Algorithm II optimizing both error, through solution. The fact part solution allows determine be selected by using threshold, long mask as approaches. Empirical results show indeed, simultaneous crucial low compared approaches reduce but do not modify filter.