Extend categorical results to Split (Inductive) Conformal Prediction and other CP variants

Establish whether the categorical commuting diagrams and morphism characterizations proven for Full Conformal Prediction extend to Inductive/Split Conformal Prediction and to other modern extensions of classical Conformal Prediction used in machine learning.

Background

The paper’s main categorical results are developed for Full (Transductive) Conformal Prediction. Many practical implementations rely on Inductive/Split CP or other variants introduced in recent years.

The authors explicitly defer determining the applicability of their categorical framework to these variants to future work, posing a concrete unknown regarding extensibility.