arXiv：cs.LG· Bj{\o}rn Leth M{\o}ller, Bulat Ibragimov, Christian Igel·· 4 小时前AI 评分34
面向忠实性信仰者：为排序相对特征重要性优化插入与删除曲线下面积
For Those Who Believe in Faithfulness: Optimizing the Area Under Insertion and Deletion Curves for Ranking Relative Feature Importance
AI 导读
研究从插入与删除曲线下面积这一忠实性度量推导出目标函数及其梯度近似方法，并建立插入曲线与 top-k 特征选择的联系，得到衡量归因质量的新型损失函数。将随机化损失与神经解释掩码框架结合得到的 Ra-NEM，可适配任意可微模型且不影响模型性能，归因在忠实性等 XAI 指标上优于其他算法，推理速度快，适合在线应用，代码已公开。
正文
Abstract:The adoption of machine learning for socially relevant tasks requires effective explainable artificial intelligence (XAI) methods to better understand the behavior of machine learning models. Attribution methods are a popular XAI approach in which input-output relationships are characterized by heat maps that reflect the relative importance of input features for a particular prediction. The quality of such maps is often assessed by measuring faithfulness based on the area under insertion and deletion curves, which measures changes in the model output as features are added and removed. In this study, we derive an objective function from this notion of faithfulness and a way to approximate its gradient. We establish the connection between insertion curves and top-$k$ feature selection, which leads to a loss function measuring the quality of attributions. Randomization of the loss allows us to efficiently approximate its gradient. To show the effectiveness of the general approach, we combine the loss function with the neural explanation mask framework. The resulting method, termed Ra-NEM, can be used with any differentiable model without affecting the model's performance. Experiments demonstrate that Ra-NEM provides accurate attributions robustly and efficiently. Compared to other algorithms, the attributions have not only higher faithfulness but also perform well in terms of other XAI metrics. The high inference speed of Ra-NEM makes the method suitable for online applications. The code is available online: this https URL
| Subjects: | Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.09844 [cs.LG] |
| (or arXiv:2610.09844v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09844 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Christian Igel [view email]
[v1]
Wed, 7 Oct 2026 11:05:08 UTC (13,447 KB)
来源：arXiv：cs.LG · arxiv.org