Adaptive logic processing in system modelling and knowledge discovery.

摘要

Recognizing the high suitability of fuzzy logic-based models for the task of system modelling and knowledge discovery, this work proposes a new approach to the automatic identification and optimization of fuzzy models. With a highly effective two-phase design process, we are able to realize adaptive logic processing in the form of structural and parametric optimization. Efficient structural learning is achieved through the use of well-established methods in Boolean minimization, with the resulting structure then refined with fuzzy neurons. The combination of a purely logic-driven architecture with this novel hybrid-learning scheme helps to develop transparent and accurate models while exhibiting superb computational efficiency. Further, this adaptive fuzzy modelling framework exhibits excellent potential for driving intelligent systems that must operate in dynamic and rapidly changing environments. In further studies, we investigate this avenue and identify various critical design issues as we propose a versatile neurofuzzy hardware platform.