Evolutionary Algorithms and Rough Sets-Based Hybrid Approach to Classificatory Decomposition of Cortical Evoked Potentials

摘要: This paper presents a novel approach to decomposition and classification of rat's cortical evoked potentials (EPs). The is based on learning sparse set basis functions using Evolutionary Algorithms (EAs). are generated in potentially overcomplete dictionary the EP components according probabilistic model data. Compared traditional, statistical signal techniques, this allows for number greater than dimensionality input signals, which can be great advantage. However, there arises an issue selecting most significant from possibly collection. especially important problems performed decomposed representation data, where only those that provide substantial discernibility between EPs different groups relevant. In paper, we propose Rough Set theory's (RS) feature selection mechanisms deal with problem. We design EA RS-based hybrid system capable and, reduced component set, classification.