AIM is a new saliency-guided adversarial feature replacement method to evaluate faithfulness of saliency maps and reliability of masking operators on image, audio, and EEG tasks.
arXiv:2202.00449
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
verdicts
UNVERDICTED 4representative citing papers
GRALIS unifies linear XAI attribution methods via a Riesz Representation Theorem-derived canonical form (Q, w, Delta), delivering seven theorems on completeness, convergence, interactions, and multi-scale extensions.
H-Sets detects higher-order feature interactions in image classifiers via Hessian-guided pair merging and attributes them with IDG-Vis to generate more interpretable saliency maps than existing marginal or coarse methods.
Current XAI methods fail because they lack well-defined problems and correctness criteria, causing them to attribute importance to features independent of the prediction target.
citing papers explorer
-
AIM: Adversarial Information Masking for Faithfulness Evaluation of Saliency Maps
AIM is a new saliency-guided adversarial feature replacement method to evaluate faithfulness of saliency maps and reliability of masking operators on image, audio, and EEG tasks.
-
GRALIS: A Unified Canonical Framework for Linear Attribution Methods via Riesz Representation
GRALIS unifies linear XAI attribution methods via a Riesz Representation Theorem-derived canonical form (Q, w, Delta), delivering seven theorems on completeness, convergence, interactions, and multi-scale extensions.
-
H-Sets: Hessian-Guided Discovery of Set-Level Feature Interactions in Image Classifiers
H-Sets detects higher-order feature interactions in image classifiers via Hessian-guided pair merging and attributes them with IDG-Vis to generate more interpretable saliency maps than existing marginal or coarse methods.
-
Explainable AI needs formalization
Current XAI methods fail because they lack well-defined problems and correctness criteria, causing them to attribute importance to features independent of the prediction target.