Scholarlink

<p>Integrating intelligence into the structure and design of the Internet of Things (IoT) has been a growing research focus, leading to the emergence of cognitive IoT (CIoT). A critical challenge in CIoT is efficiently extracting meaningful knowledge from large datasets to support diverse applications. Therefore, this research proposes a data-driven, parametrically adapted type-2 fuzzy system for knowledge discovery from massive sensory data. The sensory data is represented in an information-centric manner, where each data point carries an information value (<i>i</i>-value) that contributes to concept derivation. By applying k-means clustering to <i>i</i>-values, the proposed fuzzy system effectively identifies significant concepts (patterns). Experimental evaluation of massive environmental and healthcare data demonstrates the method’s effectiveness over competing approaches, with uMbRAE, RMSE, MAPE, and squared-MAPE all remaining below one. Furthermore, this approach functions as a stream reasoner, enabling real-time concept extraction from large-scale data in CIoT applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Data-driven parametric adaption in type-2 fuzzy logic for significant pattern discovery using massive heterogeneous data in cognitive IoT

  • 期刊论文

  • 发表日期：

    2025年6月2日

查看全文
  • Vidyapati Jha,
  • Priyanka Tripathi

摘要

Integrating intelligence into the structure and design of the Internet of Things (IoT) has been a growing research focus, leading to the emergence of cognitive IoT (CIoT). A critical challenge in CIoT is efficiently extracting meaningful knowledge from large datasets to support diverse applications. Therefore, this research proposes a data-driven, parametrically adapted type-2 fuzzy system for knowledge discovery from massive sensory data. The sensory data is represented in an information-centric manner, where each data point carries an information value (i-value) that contributes to concept derivation. By applying k-means clustering to i-values, the proposed fuzzy system effectively identifies significant concepts (patterns). Experimental evaluation of massive environmental and healthcare data demonstrates the method’s effectiveness over competing approaches, with uMbRAE, RMSE, MAPE, and squared-MAPE all remaining below one. Furthermore, this approach functions as a stream reasoner, enabling real-time concept extraction from large-scale data in CIoT applications.