Scholarlink

In this work, we discuss a possible extension of the recently developed Conformal Association Rule Mining (CARM) technique to address multiple errors in the data. We utilize the calibration properties of the Conformal Prediction framework for reliable machine learning to bound the false alarm rate and introduce a flexible stopping criterion for the iterations. We developed new non-conformity measures for this task and provided a case study with images of handwritten digits.

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

Conformal Mining for Multiple Error Correction in the Data

  • 章节

  • 发表日期：

    2026年

查看全文
  • Ilia Nouretdinov

摘要

In this work, we discuss a possible extension of the recently developed Conformal Association Rule Mining (CARM) technique to address multiple errors in the data. We utilize the calibration properties of the Conformal Prediction framework for reliable machine learning to bound the false alarm rate and introduce a flexible stopping criterion for the iterations. We developed new non-conformity measures for this task and provided a case study with images of handwritten digits.