Scholarlink

<p>A fuzzy rule-based system with the Generalized Modus Ponens (GMP) requires choosing the inference mechanism and its operators. These parameters affect the output of the system. In this paper, we aim to select the best possible output with respect to an input from a set of inference methods for solving the GMP through a decision-making system. To develop this system, we propose a fuzzy rule comparison measure that returns the similarity degree between two fuzzy <i>if-then</i> rules, in which the influence of both antecedent and consequent fuzzy sets is considered. To construct such comparison measures, we develop some generalizations of the Jaccard index based on grouping and overlap indices. Finally, we illustrate our decision-making system in a gastric cancer survival length prediction problem by evaluating various inference configurations across distinct clinical patient profiles and applying the proposed method to select the best one. Our results show that the generalized Jaccard index performs better than the classical Jaccard index at identifying similarities between fuzzy rules. Moreover, our selection approach effectively identifies the best inference method, in this case study, a specific configuration within the interpolation-based methods.</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

Comparison measures of fuzzy inference rules based on generalizations of jaccard indices applied to inference method selection

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月3日

查看全文
  • Urio-Larrea Asier,
  • Zdenko Takáč,
  • Graçaliz Pereira Dimuro,
  • Javier Andreu-Pérez,
  • Giancarlo Lucca,
  • Cedric Marco-Detchart,
  • Anderson Cruz,
  • Heloisa Camargo,
  • Humberto Bustince

摘要

A fuzzy rule-based system with the Generalized Modus Ponens (GMP) requires choosing the inference mechanism and its operators. These parameters affect the output of the system. In this paper, we aim to select the best possible output with respect to an input from a set of inference methods for solving the GMP through a decision-making system. To develop this system, we propose a fuzzy rule comparison measure that returns the similarity degree between two fuzzy if-then rules, in which the influence of both antecedent and consequent fuzzy sets is considered. To construct such comparison measures, we develop some generalizations of the Jaccard index based on grouping and overlap indices. Finally, we illustrate our decision-making system in a gastric cancer survival length prediction problem by evaluating various inference configurations across distinct clinical patient profiles and applying the proposed method to select the best one. Our results show that the generalized Jaccard index performs better than the classical Jaccard index at identifying similarities between fuzzy rules. Moreover, our selection approach effectively identifies the best inference method, in this case study, a specific configuration within the interpolation-based methods.