Fuzzy interpolation systems with knowledge extraction and adaptation

摘要: Fuzzy inference system provides an effective means for representing and processing vagueness imprecision. Conventional fuzzy modelling requires either complete experts’ knowledge or given datasets to generate rule bases such that the input spaces can be fully covered. Although interpolation enhances power of conventional approaches by addressing problem lack represented in bases, it is still difficult real-world applications obtain sufficient and/or data a sparse base support interpolation. Also, generated are usually fixed therefore cannot dynamic situations. In addition, all existing were developed based on Mamdani model, which not applicable TSK model. It significantly restricts applicability systems. This PhD work, first part, presents novel approach, termed “TSK+ inference”, address issue performing over bases. The proposed TSK+ approach extends considering degree similarity between inputs corresponding antecedents instead overlapped match degree, allows performed dense imbalanced order data-driven generation method also presented this work. has been further extended deal with interval type-2 sets. effectiveness enhancing demonstrated through two applications: network intrusion detection system, quality service management system. In second part new adaptation relax requirement generation, minimal even without priori knowledge. mimics pedagogic experiential learning, achieves automatic transferring proceeding performance experiences when inferences. evaluated only mathematical model but well-known control problem, inverted pendulum. experimental results show running, thus demonstrating approach.