BOHM extracts multi-resolution attribution trees from existing routing weights in hierarchical AI systems, providing zero-cost explanations that correlate with SHAP when routing is near-optimal.
A unified approach to interpreting model predictions
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3representative citing papers
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.
Quadrature-TreeSHAP computes Shapley values and higher-order interactions for tree ensembles via 8-point Gauss-Legendre quadrature on a Banzhaf polynomial, removing depth dependence while reaching machine precision and delivering speedups up to 1200x for interactions.
citing papers explorer
-
BOHM: Zero-Cost Hierarchical Attribution for Compound AI Systems
BOHM extracts multi-resolution attribution trees from existing routing weights in hierarchical AI systems, providing zero-cost explanations that correlate with SHAP when routing is near-optimal.
-
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.
-
Quadrature-TreeSHAP: Depth-Independent TreeSHAP and Shapley Interactions
Quadrature-TreeSHAP computes Shapley values and higher-order interactions for tree ensembles via 8-point Gauss-Legendre quadrature on a Banzhaf polynomial, removing depth dependence while reaching machine precision and delivering speedups up to 1200x for interactions.