Scholarlink

<p>Concept-cognitive learning has made significant strides in cognitive science by simulating human brain processes, especially in handling fuzzy data. Despite recent achievements, challenges persist in achieving comprehensive and accurate cognition in fuzzy contexts, reflected in: insufficient factual precision in pseudo-concept spaces, restricted scalability of generalized concepts, and subjectivity arising from excessive parameter reliance. To address these issues, we develop two innovative approaches and its simplified variant by integrating granular-ball computing theory into concept-cognitive learning. In this paper, we introduce a concept refine method that captures nuanced characteristics enhancing the completeness and realism of cognition. Additionally, leveraging the Gaussian kernel, a novel similarity degree is defined to recognize new clues and design a concept-cognitive algorithm based on the granular-ball splitting subsequently. To optimize memory usage in GBCCL, we present the concept of the ‘pseudo-granular-ball’ and introduce a streamlined model. Experimental results demonstrate the effectiveness, facticity, and objectivity of our proposed methods in achieving cognitive learning within fuzzy contexts. The code is available at <a href="https://github.com/KouYinan/GBCCL.git.">https://github.com/KouYinan/GBCCL.git.</a></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

A novel self-adaptive fuzzy concept-cognitive learning based granular-ball splitting

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Qijun Wang,
  • Yidong Lin,
  • Jinjin Li,
  • Menglei Lin,
  • Taoju Liang

摘要

Concept-cognitive learning has made significant strides in cognitive science by simulating human brain processes, especially in handling fuzzy data. Despite recent achievements, challenges persist in achieving comprehensive and accurate cognition in fuzzy contexts, reflected in: insufficient factual precision in pseudo-concept spaces, restricted scalability of generalized concepts, and subjectivity arising from excessive parameter reliance. To address these issues, we develop two innovative approaches and its simplified variant by integrating granular-ball computing theory into concept-cognitive learning. In this paper, we introduce a concept refine method that captures nuanced characteristics enhancing the completeness and realism of cognition. Additionally, leveraging the Gaussian kernel, a novel similarity degree is defined to recognize new clues and design a concept-cognitive algorithm based on the granular-ball splitting subsequently. To optimize memory usage in GBCCL, we present the concept of the ‘pseudo-granular-ball’ and introduce a streamlined model. Experimental results demonstrate the effectiveness, facticity, and objectivity of our proposed methods in achieving cognitive learning within fuzzy contexts. The code is available at https://github.com/KouYinan/GBCCL.git.