Saved in:
Bibliographic Details
Main Authors: Zhang, Siyu, Mcmillan, Kenneth
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.04588
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918230662053888
author Zhang, Siyu
Mcmillan, Kenneth
author_facet Zhang, Siyu
Mcmillan, Kenneth
contents Faithfulness metrics such as insertion and deletion evaluate how feature removal affects model outputs but overlook whether explanations preserve the computational pathway the network actually uses. We show that external metrics can be maximized through alternative pathways -- perturbations that reroute computation via different feature detectors while preserving output behavior. To address this, we propose activation preservation as a tractable proxy for preserving computational pathways We introduce Faithfulness-guided Ensemble Interpretation (FEI), which jointly optimizes external faithfulness (via ensemble quantile optimization of insertion/deletion curves) and internal faithfulness (via selective gradient clipping). Across VGG and ResNet on ImageNet and CUB-200-2011, FEI achieves state-of-the-art insertion/deletion scores while maintaining significantly lower activation deviation, showing that both external and internal faithfulness are essential for reliable explanations.
format Preprint
id arxiv_https___arxiv_org_abs_2509_04588
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Output Faithfulness: Learning Attributions that Preserve Computational Pathways
Zhang, Siyu
Mcmillan, Kenneth
Machine Learning
Artificial Intelligence
Faithfulness metrics such as insertion and deletion evaluate how feature removal affects model outputs but overlook whether explanations preserve the computational pathway the network actually uses. We show that external metrics can be maximized through alternative pathways -- perturbations that reroute computation via different feature detectors while preserving output behavior. To address this, we propose activation preservation as a tractable proxy for preserving computational pathways We introduce Faithfulness-guided Ensemble Interpretation (FEI), which jointly optimizes external faithfulness (via ensemble quantile optimization of insertion/deletion curves) and internal faithfulness (via selective gradient clipping). Across VGG and ResNet on ImageNet and CUB-200-2011, FEI achieves state-of-the-art insertion/deletion scores while maintaining significantly lower activation deviation, showing that both external and internal faithfulness are essential for reliable explanations.
title Beyond Output Faithfulness: Learning Attributions that Preserve Computational Pathways
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2509.04588