Scholarlink

<p>Machine learning (ML) has emerged as a transformative tool in concrete technology, allowing precise prediction, optimization, and decision-making across diverse materials and applications. Despite the predictive power of this tool, ML models are often complex and difficult to interpret, which inspired the increased usage of explainable and interpretable techniques. This study presents a comprehensive scientometric and systematic review of explainable ML applications in concrete research. The scientometric analysis reveals a steady growth of publications signifying global interest in explainable ML. The keyword co-occurrence mapping indicates that techniques such as extreme gradient boosting and random forests are most frequently applied, with studies addressing concrete strength, durability, and sustainability-related properties. SHapley Additive exPlanations (SHAP) is identified as the main method for model interpretation. The review further highlights data-related challenges, including dataset availability, heterogeneity across concrete types, and insufficient reporting, which underscore the need for rigorous data handling and model validation. This study provides a structured roadmap for advancing explainable ML in material-level scope of concrete technology and advance sustainable, resilient, and reliable infrastructure design.</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

Explainable Machine Learning in Concrete Technology: A State-of-the-Art Review with Future Innovations

  • 期刊论文

  • 发表日期：

    2026年1月29日

查看全文
  • Ibrahim A. Tijani,
  • Tadesse G. Wakjira,
  • Dima Kanaan,
  • M. Shahria Alam,
  • Hasan Haroglu

摘要

Machine learning (ML) has emerged as a transformative tool in concrete technology, allowing precise prediction, optimization, and decision-making across diverse materials and applications. Despite the predictive power of this tool, ML models are often complex and difficult to interpret, which inspired the increased usage of explainable and interpretable techniques. This study presents a comprehensive scientometric and systematic review of explainable ML applications in concrete research. The scientometric analysis reveals a steady growth of publications signifying global interest in explainable ML. The keyword co-occurrence mapping indicates that techniques such as extreme gradient boosting and random forests are most frequently applied, with studies addressing concrete strength, durability, and sustainability-related properties. SHapley Additive exPlanations (SHAP) is identified as the main method for model interpretation. The review further highlights data-related challenges, including dataset availability, heterogeneity across concrete types, and insufficient reporting, which underscore the need for rigorous data handling and model validation. This study provides a structured roadmap for advancing explainable ML in material-level scope of concrete technology and advance sustainable, resilient, and reliable infrastructure design.