A self-adaptive fuzzy learning system for streaming data prediction

In this paper, a novel self-adaptive fuzzy learning (SAFL) system is proposed for streaming data prediction. SAFL self-learns from streams predictive model composed of set prototype-based rules, with each which representing certain local distribution, and continuously self-evolves to follow the changing patterns in non-stationary environments. Unlike conventional evolving systems, both inference consequent parameter schemes utilised by are simplified so that only small number selected rules within rule base involved output generation updating during cycle. Such simplification not significantly reduces system’s computational complexity but also increases its prediction precision. addition, theoretical empirical investigations guarantee stability resulting SAFL. Comparative experimental studies on wide variety benchmark real-world problems demonstrate able learn highly efficient manner make predictions great accuracy, revealing effectiveness validity approach.