Fuzzy Rough Theory And Neural Network Applied In Information Handling

In this paper, based on fuzzy set, rough set and neural network theory, several different combines are developing intelligent inference handling system to demonstrate the effectiveness of information analysis and inference techniques that utilize fuzzy logic, neural network, rough set theory and genetic algorithms. Firstly, the classical fuzzy set, rough sets, GA algorithm and neural network theory are introduced. The later, data reduction based on attribute membership grades is used to deal with continuous data which can't be reduced by classical rough set. A rough adaptive neural fuzzy inference system is introduced. A rough neuron consists of an upper and a lower neuron. Rough neurons can be used to effectively represent an interval or a set of values; a genetic neural fuzzy inference system is presented, Genetic algorithm is used to optimize the structure of the system and the membership function of each fuzzy term because of its capability of parallel and global search, and we present an approach to obtain a reduced genetic neural fuzzy system. The reduction is carried out though an iterative algorithm aiming at selecting a minimal number of rules of the model. To decide which rules we may eliminate, dependency in rough set theory is used. Finally, a rough-fuzzy learning algorithm is developed for rough adaptive neural fuzzy inference system, and GA-rough algorithm is developed for GA-rough neural fuzzy inference system. The results predicted by the rough adaptive neural fuzzy inference system show it can perform very well in dealing with the electrocardiogram which has two outputs; And data reduction based on attribute membership grades is well done in reducing continuous data; The GA-rough algorithm can fast optimize the neural fuzzy inference system, and which is applied in decision-classify and function approach.