REVIEW 9 cited by
On quantitative aspects of model interpretability
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Despite the growing body of work in interpretable machine learning, it remains unclear how to evaluate different explainability methods without resorting to qualitative assessment and user-studies. While interpretability is an inherently subjective matter, previous works in cognitive science and epistemology have shown that good explanations do possess aspects that can be objectively judged apart from fidelity), such assimplicity and broadness. In this paper we propose a set of metrics to programmatically evaluate interpretability methods along these dimensions. In particular, we argue that the performance of methods along these dimensions can be orthogonally imputed to two conceptual parts, namely the feature extractor and the actual explainability method. We experimentally validate our metrics on different benchmark tasks and show how they can be used to guide a practitioner in the selection of the most appropriate method for the task at hand.
Forward citations
Cited by 9 Pith papers
-
Value bounds and Convergence Analysis for Averages of LRP attributions
Averaged LRP-beta attributions have Hoeffding convergence bounds independent of weight norms, unlike gradient-based explanations.
-
DeepFaith: A Domain-Free and Model-Agnostic Unified Framework for Highly Faithful Explanations
A single trained explainer that unifies multiple faithfulness metrics and generates model-agnostic explanations with high measured faithfulness.
-
Reconciling Privacy and Explainability in High-Stakes: A Systematic Inquiry
Gradient-based explainers yield almost uncorrelated attributions on DP-trained chest X-ray models, so the authors recommend privatizing explanations from a non-private model instead.
-
VARSHAP: Addressing Global Dependency Problems in Explainable AI with Variance-Based Local Feature Attribution
VARSHAP defines local feature attribution as the Shapley value of a variance-reduction game and claims greater stability than SHAP and LIME.
-
Advancing Attribution-Based Neural Network Explainability through Relative Absolute Magnitude Layer-Wise Relevance Propagation and Multi-Component Evaluation
A new LRP rule that scales attributions by absolute activation magnitude, plus a unified evaluation metric, is tested across three architectures and two datasets.
-
From Flexibility to Manipulation: The Slippery Slope of XAI Evaluation
XAI faithfulness evaluation can be flipped by changing perturbation type, partition size, or normalization, allowing a chosen explanation method to win, with a ranking-based mitigation called Mean Resilience Rank.
-
Explainability for Vision Foundation Models: A Survey
A structured review of 122 papers on explainability for vision foundation models, with a taxonomy and the finding that quantitative evaluation is rare (36%).
-
Establishing and Evaluating Trustworthy AI: Overview and Research Challenges
A semi-structured literature review synthesizing six trustworthy AI requirements and their evaluation methods, plus cross-cutting research challenges.
-
Unlocking the Black Box: Analysing the EU Artificial Intelligence Act's Framework for Explainability in AI
A legal review arguing that explainability is central to the EU AI Act but is not yet translated into concrete technical and legal standards.
Discussion (0). Continue with ORCID to comment.