Adaptive Multi codebook Fuzzy Neuro Generalized Learning Vector Quantization for sleep stages classification

In this paper, a new codebook based learning method, Adaptive Multicodebook Fuzzy Neuro Generalized Learning Vector Quantization (FNGLVQ), is proposed. The main contribution of this paper is the use of multi codebook which is adaptive in nature to the distribution of the data. The number and position of the codebook is determined through clustering approach. In this research, a decision tree based clustering, CLTree, is used to cluster the data to get the initial placement of the codebook. The advantage of using CLTree against other clustering method is CLTree do not need the number of cluster as initial input. In average, this method improves the accuracy rate of Mitra data 3 and 4 class 2% and 2.12%, respectively compared to the single codebook approach.