RuleSHAP, which combines global SHAP values with XGBoost and LASSO rule selection, recovers injected univariate, conjunctive, and non-convex LLM behavior triggers more faithfully than RuleFit and SHAP alone.
The role of artificial intelligence in achieving the sustainable development goals.Nature communications, 11(1):1–10, 2020
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Can Global XAI Methods Reveal Injected Behaviours in LLMs? SHAP vs Rule Extraction vs RuleSHAP
RuleSHAP, which combines global SHAP values with XGBoost and LASSO rule selection, recovers injected univariate, conjunctive, and non-convex LLM behavior triggers more faithfully than RuleFit and SHAP alone.