ENCP: Episode-Normalized Conformal Prediction for Vision-and-Language Navigation
ENCP calibrates uncertainty over whole navigation episodes, not isolated steps.
The paper argues that standard conformal prediction breaks down for vision-and-language navigation because an agent’s decisions are sequential, dependent, and variable in length. ENCP rescales nonconformity scores by residual policy confidence and calibrates the maximum score per episode. Under exchangeable calibration and test episodes, the authors claim step-level ground-truth coverage of at least \(1-\alpha\). They report that ENCP hit all empirical step-coverage targets across four VLN policies and three scoring methods on R2R and REVERIE. ArXiv · AI/CL/LG's note
The paper argues that standard conformal prediction breaks down for vision-and-language navigation because an agent’s decisions are sequential, dependent, and variable in length. ENCP rescales nonconformity scores by residual policy confidence and calibrates the maximum score per episode. Under exchangeable calibration and test episodes, the authors claim step-level ground-truth coverage of at least \(1-\alpha\). They report that ENCP hit all empirical step-coverage targets across four VLN policies and three scoring methods on R2R and REVERIE. ArXiv · AI/CL/LG's note
score 4