NeuFuz: the integration of fuzzy logic and neural nets offers designers tremendous application advantage - today

A new technology combines the best features of two forefront technologies-neural net and fuzzy logic-into a hybrid that eliminates the drawbacks of each singular technology. Dubbed "NeuFuz" this technology relies on neural nets to learn a systems's behavior using system input-output data, then automatically generates fuzzy rules and membership functions from the learned system knowledge. Once an optimized solution is derived, NeuFuz converts it into either standard C code or assembly language code for an embedded processor. NeuFuz can provide significant advantages in complex applications such as object, pattern or voice recognition; language processing, traffic management and even weather forecasting. The technology reduces design time and significantly improves system performance, accuracy and reliability. A fuzzy rule verifier verifies the solution and optimizes the number of rules and membership functions. The adaptation capability of NeuFuz helps it learn from experience and change fuzzy rules and membership functions accordingly. This paper discusses the benefits of the NeuFuz technology, its implementation and the types of applications.