域偏移下行人检测的置信度控制XAI审计
原标题：Confidence-Controlled XAI Auditing for Pedestrian Detection under Domain Shift
AI 摘要
该研究对固定YOLOv8s行人检测器在PIE和JAAD两个驾驶数据集上的事后解释进行审计，采用基于ROI的D-Deletion、冻结置信度三分位、秩检验、bootstrap区间和Holm校正。结果显示，基于删除的忠实度与解释时的检测强度强耦合，Spearman相关性使按置信度分层的朴素比较不可靠；在固定区间内控制检测强度后，D-RISE忠实度在中心范围仍具域依赖性，PIE的D-Deletion高于JAAD且经Holm校正显著。非扰动EigenCAM基线忠实度低于D-RISE但也表现出分数耦合，表明该效应与删除式评估设置有关，作者据此提出在域偏移下进行置信度控制的XAI审计。
正文节选
Confidence-Controlled XAI Auditing for Pedestrian Detection under Domain Shift Abstract Explainability is increasingly required for perception models in intelligent vehicles, yet whether explanations remain faithful under driving domain shift is still poorly understood. This work audits post-hoc explanations of a fixed YOLOv8s pedestrian detector across PIE and JAAD using ROI-based D-Deletion, frozen confidence terciles, rank-based tests, bootstrap intervals, and Holm correction. The audit shows