Fuzzy min-max neural network based classification of underwater layered media due to attenuation effects

Summary form only given. A simple nonlinear structure for the implementation of underwater layered media modeling based on time delay in layers and exponential decay of the output signals´ amplitude due to attenuation effects is proposed. The results of computerized simulation of this model are presented. A feed-forward fuzzy neural network classifier that uses min-max amplitude ranges to define classes is designed and evaluated based on a computer simulation of synthetic data. A supervised fuzzy min-max learning rule is described and employed to update the weights corresponding to the input data sets used for training. Upon completion of the training stage, the classifier is tested for different numbers of test data sets and it would be able to tell the class of the tested data. Each node of the output layer of the classifier represents a class and recognizes the grade of minimum or maximum values of the amplitudes. This means that during recall each class node produces an output value that represents the degree to which the input pattern fits within the represented class. Classifying the peak point of the amplitudes allows the identification of different layers of the media. Computer simulation results are used in conjunction with the proposed fuzzy min-max neural network. The results suggest that the application of fuzzy min-max neural networks in pattern recognition will enable automatic classification of the data collections and clusters. In this paper, the relationship between fuzzy sets and pattern classification, the fuzzy min-max neural network model implementation, learning and recalling algorithms, and results of the modeling are given. The proposed fuzzy min-max neural network is demonstrated an ability to classify the layered media data sets, and it will be able to distinguish between their min and max points