Scholarlink

This paper presents a semantically coherent framework for integrating reasoning and learning through Sugeno–Weber consistent connective systems. It introduces a method for translating first-order logic rules into differentiable loss functions by directly applying Sugeno-Weber generators to the truth degrees of ground atoms. The proposed framework integrates symbolic reasoning with gradient-based learning, ensuring both semantic precision and a solid theoretical foundation. Experimental results demonstrate that the proposed approach achieves faster convergence, emphasizing the utility of Sugeno–Weber connectives for neuro-symbolic integration while maintaining logical consistency and differentiability.

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

Inspired-Logic Loss Functions for Deep Learning Using Sugeno-Weber Connective Systems

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mohamed Hssini,
  • Mustapha Atraoui

摘要

This paper presents a semantically coherent framework for integrating reasoning and learning through Sugeno–Weber consistent connective systems. It introduces a method for translating first-order logic rules into differentiable loss functions by directly applying Sugeno-Weber generators to the truth degrees of ground atoms. The proposed framework integrates symbolic reasoning with gradient-based learning, ensuring both semantic precision and a solid theoretical foundation. Experimental results demonstrate that the proposed approach achieves faster convergence, emphasizing the utility of Sugeno–Weber connectives for neuro-symbolic integration while maintaining logical consistency and differentiability.