发表机构
Faculty of Engineering, University of Porto; Artificial Intelligence and Computer Science Lab (LIACC); Fraunhofer AICOS Portugal; BrightFactory; LIACS, Leiden University; School of Computing and Mathematical Sciences, University of Waikato（波尔图大学工程学院; 人工智能与计算机科学实验室（LIACC）; 弗劳恩霍夫葡萄牙人工智能与计算机科学中心; 光明工厂; 莱顿大学LIACS; 怀卡托大学计算与数学科学学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出罗生门对齐（RA）评估模型功能相似性，引入几何视角，提出几何RA，通过90多个数据集实验分析，表明几何与分布对齐提供不同互补视角，RA可用于模型选择等多用途。
AI 中文摘要
我们提出了罗生门对齐（RA），一种评估两个模型之间功能相似性的新方法。现有的功能相似性度量是基于分布的，量化应用于实际数据的模型输出之间的差异。然而，这些度量仅对可用数据所代表的输入空间区域具有生态有效性。我们引入了一种关于功能模型相似性的几何视角，它能在整个数据空间中进行估计，提供独立于任何特定数据分布的决策边界对齐的全面视图。我们还提出了几何罗生门对齐作为几何相似性的度量，它使用从实例空间均匀采样的数据来计算。我们对90多个数据集进行了实验分析，研究了模型对齐与预测准确性不同的关键情况。我们的结果表明，几何对齐和分布对齐为模型和算法之间的相似性提供了不同且互补的视角。RA可用于多种目的，包括模型选择、集成构建以及增强机器学习模型和算法的可解释性。
英文摘要
We propose Rashomon Alignment (RA), a new measure to assess functional similarity between two models. Existing functional similarity measures are distributional, quantifying differences between outputs of models applied to real-world data. However, these measures can be regarded as ecologically valid only for regions in the input space represented by the available data. We introduce a geometrical perspective on functional model similarity, which estimates it across the entire data space, offering a comprehensive view of decision boundary alignment independent of any specific data distribution. We also propose geometric Rashomon Alignment as a measure of geometrical similarity, which is computed using data uniformly sampled from the instance space. We perform an experimental analysis on more than 90 datasets, examining critical cases where model alignment diverges from predictive accuracy. Our results show that geometrical and distributional alignment provide different and complementary perspectives on the similarity between models and algorithms. RA can be used for multiple purposes, including model selection, ensemble construction, and enhanced interpretability of machine learning models and algorithms.
