Scholarlink

<p>A novel model integrating TabPFN and SHAP was developed to accurately, efficiently, and interpretably predict slag viscosity. Unlike traditional machine learning models, TabPFN model achieves high accuracy (R<sup>2</sup> = 0.9831, RMSE = 1.4521 Pa·s, MAE = 0.6004 dPa·s, hit ratio = 94.25 pct) with fast response (0.26 s) and no extensive hyperparameter tuning. SHAP reveals global and local factor influences. Meanwhile, real-time updating software was developed, enabling model optimization without extensive hyperparameter tuning as new data are introduced.</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

TabPFN-SHAP-Based Slag Viscosity Prediction Model with High Accuracy, Efficiency, and Interpretability

  • 期刊论文

  • 发表日期：

    2025年7月7日

查看全文
  • Zi-cheng Xin,
  • Jiang-shan Zhang,
  • Mo Lan,
  • Qing Liu

摘要

A novel model integrating TabPFN and SHAP was developed to accurately, efficiently, and interpretably predict slag viscosity. Unlike traditional machine learning models, TabPFN model achieves high accuracy (R2 = 0.9831, RMSE = 1.4521 Pa·s, MAE = 0.6004 dPa·s, hit ratio = 94.25 pct) with fast response (0.26 s) and no extensive hyperparameter tuning. SHAP reveals global and local factor influences. Meanwhile, real-time updating software was developed, enabling model optimization without extensive hyperparameter tuning as new data are introduced.