Reference · The Applied Higher-Order Functions catalogue

rainbow.algorithms.proofs.ahof.ensemble-disagreement

The Applied Higher-Order Functions cataloguesrc/rainbow/algorithms/proofs/ahof/ensemble_disagreement.clj1 definitions, 3 theorems

AHoF V61 Ensemble Disagreement: quantify predictive uncertainty by the spread of an ensemble's predictions and flag an input as uncertain when the disagreement reaches a threshold: flag-uncertain? disagreement threshold = decide (threshold <= disagreement). A disagreement of 3/4 against threshold 1/2 is flagged uncertain; a disagreement of 1/4 is confident (not flagged). Proved native and axiom-free (decision, flagged and confident instances, at-most certificate). Requires audit/decide-nnrat-at-most, nnrat/at-most, nnrat/decide-at-most-true-sound (all in the D63 checkpoint). Scope (measured by the kernel): this module has 3 worked instances (statements about specific values). None of its theorems is a guarantee about the algorithm for all inputs; `lein verify rainbow.algorithms.proofs.ahof.ensemble-disagreement` prints this classification.

ensemble/flag-uncertain? :
  (x₁ x₂ : ℚ≥0)
  → data/Bool

ensemble/high-disagreement-flagged :
  → bo/is-safe 3/4 1/2 = data/true

ensemble/disagreement-certificate :
  → 1/2 ≤ 3/4

ensemble/low-disagreement-confident :
  → bo/is-safe 1/4 1/2 = data/false