Preprint
Aug 2026
Judge, Retrieve, or Abstain: Uncertainty-Guarded LLM Judging with Provable Risk Guarantees
A risk-controlled framework that calibrates uncertainty thresholds on a held-out set so that the false discovery rate among accepted verdicts remains below a user-specified level~$\alpha$ with high probability, using finite-sample Clopper--Pearson intervals is proposed.
Sher Badshah, Ali Emami, Hassan Sajjad
· 1 citation