Sanity Checks for Saliency Maps¶
Adebayo, J., Gilmer, Justin, Muelly, Michael, Goodfellow, Ian, et al. (2018). Sanity Checks for Saliency Maps. Advances in Neural Information Processing Systems 31.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Mechanisms¶
- Salience Map or Attention Heatmap
- Its failure mode is superficial-cue lock-in: a salient-but-irrelevant hotspot captures attention and biases everything that follows, and a heatmap can look explanatory without being faithful — saliency maps can highlight spurious correlations rather than the true cause.
This sourceShows that some saliency maps can look similar even after model parameters or training labels are randomized, undermining their fidelity as explanations.
- Its failure mode is superficial-cue lock-in: a salient-but-irrelevant hotspot captures attention and biases everything that follows, and a heatmap can look explanatory without being faithful — saliency maps can highlight spurious correlations rather than the true cause.
Verification¶
Does it exist? Not checked yet. This entry carries no identifier to resolve. It was extracted from the citation as written in the article, normalized, and deduplicated against the rest of the registry.
Does it back the claim? Not recorded. The single citation of this work carries no recorded support check.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:126e30f5edba · see in the full table