Apr 27
H-Sets: Hessian-Guided Discovery of Set-Level Feature Interactions in Image Classifiers
★★★★★
significance 2/5
Researchers introduce H-Sets, a new framework for discovering and attributing higher-order feature interactions in image classifiers. The method uses input Hessians and a two-stage process to create more faithful and sparse saliency maps compared to existing attribution methods.
Why it matters
Improved attribution fidelity is essential for debugging the opaque, higher-order feature interactions driving modern computer vision models.
Tags
#feature attribution #interpretability #image classification #hessian-guided #deep learningRelated coverage
- Global South OpportunitiesPivotal Research Fellowship 2026 (Q3): AI Safety Research Opportunity - Global South Opportunities
- arXiv cs.AIAn Intelligent Fault Diagnosis Method for General Aviation Aircraft Based on Multi-Fidelity Digital Twin and FMEA Knowledge Enhancement
- arXiv cs.AIPExA: Parallel Exploration Agent for Complex Text-to-SQL
- arXiv cs.AIThe Power of Power Law: Asymmetry Enables Compositional Reasoning
- arXiv cs.AIOn the Existence of an Inverse Solution for Preference-Based Reductions in Argumentation