Ternary Decision Trees with Locally-Adaptive Uncertainty Zones

William Smits · 2026

Read Ternary Decision Trees with Locally-Adaptive Uncertainty Zones by William Smits (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-22740.txt.

Description

Smits introduces ternary decision trees, which add an uncertainty zone around each split so near threshold cases avoid false confidence. A decision theoretic framework fixes the zone width from node local costs and proves four properties, including accuracy decomposition and asymptotic consistency of the margin method. Cases inside the zone blend both child subtrees and carry an uncertain flag. Across most OpenML-CC18 data sets, all five zone width estimators with probabilistic routing beat standard CART on decided accuracy, with the margin method best and no tuning knobs.

Catalog record

CREATORWilliam Smits
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE64018
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.22740

Browse the full library · About the Tower of Babel