Self-Organizing Fuzzy Belief Inference System for Classification

Evolving fuzzy systems (EFSs) are widely known as a powerful tool for streaming data prediction. In this article, novel zero-order EFS with unique belief structure is proposed stream classification. Thanks to new structure, the model can handle interclass overlaps in natural way and better capture underlying multimodel of streams form prototypes. Utilizing data-driven soft thresholds, self-organizes set prototype-based if – xmlns:xlink="http://www.w3.org/1999/xlink">then rules from classification, its learning outcomes practically meaningful. With no requirement prior knowledge problem domain, capable self-determining appropriate level granularity rule-based construction, while enabling users specify their preferences on degree fineness base. Numerical examples demonstrate superior performance wide range stationary nonstationary classification benchmark problems.