An explainable semi-supervised self-organizing fuzzy inference system for streaming data classification

As a powerful tool for data streams processing, the vast majority of existing evolving intelligent systems (EISs) learn prediction models from in supervised manner. However, high-quality labelled can be difficult to obtain many real-world classification applications concerning streams, though unlabelled is plentiful. To overcome labelling bottleneck and construct stronger model, novel semi-supervised EIS proposed this paper. After being primed with small amount data, method capable continuously self-developing its system structure self-updating meta-parameters chunk-by-chunk non-iterative, exploratory manner by exploiting pseudo-labelling strategy. Thanks transparent prototype-based human-understandable reasoning process, provide users high explainability interpretability while achieving great precision. Experimental investigation demonstrates superior performance method.