Skip to content
Back to skills

Auditing Visualizations Transparency Methods Struggle To Detect Anomalous Behavior

ASecurity

**arXiv ID:** 2206.13498 **Authors:** Jean-Stanislas Denain, Jacob Steinhardt **Published:** 2022-06-27T17:59:08Z **Abstract:** Model visualizations provide information that outputs alone might miss. But can we trust that model visualizations reflect model behavior? For instance, can they diagnose abnormal behavior such as planted backdoors or overregularization? To evaluate visualization methods, we test whether they assign different visualizations to anomalously trained models and normal mo...

  • 3 stars
  • 0 votes
  • 0 copies
  • 6 views
  • Added September 11, 2026
toolsrust

Security analysis

A100/100

Scanned September 11, 2026

npx -y skills add hiyenwong/ai_collection --skill auditing-visualizations-transparency-methods-struggle-to-detect-anomalous-behavior --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Auditing Visualizations Transparency Methods Struggle To Detect Anomalous Behavior?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Auditing Visualizations Transparency Methods Struggle To Detect Anomalous Behavior
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hiyenwong-auditing-visualizations-transparency-methods-strug/badge)](https://www.skillsdirectory.com/skills/hiyenwong-auditing-visualizations-transparency-methods-strug)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
# Auditing Visualizations: Transparency Methods Struggle to Detect Anomalous Behavior

**arXiv ID:** 2206.13498
**Authors:** Jean-Stanislas Denain, Jacob Steinhardt
**Published:** 2022-06-27T17:59:08Z
**Abstract:**
Model visualizations provide information that outputs alone might miss. But can we trust that model visualizations reflect model behavior? For instance, can they diagnose abnormal behavior such as planted backdoors or overregularization? To evaluate visualization methods, we test whether they assign different visualizations to anomalously trained models and normal models. We find that while existing methods can detect models with starkly anomalous behavior, they struggle to identify more subtle anomalies. Moreover, they often fail to recognize the inputs that induce anomalous behavior, e.g. images containing a spurious cue. These results reveal blind spots and limitations of some popular model visualizations. By introducing a novel evaluation framework for visualizations, our work paves the way for developing more reliable model transparency methods in the future.

## Skill Description

This skill is generated from the arXiv paper: Auditing Visualizations: Transparency Methods Struggle to Detect Anomalous Behavior (2206.13498).

## How to Use

[To be filled in by the user or by future automation]

## References

- [arXiv:2206.13498](http://arxiv.org/abs/2206.13498v2)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in tools →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.