Paper 03BranchVersion 2
Learning Effective Interfaces from Opaque Stochastic Systems: Capacity, Selection, and Validation Limits
Authority role
Learning effective interfaces: the gap between capacity, candidate production, selection and validation.
Summary
A shared-data comparison on 48 opaque synthetic systems evaluates complete interaction laws, including joint outputs, delayed effects and resource sensitivity. The primary model portfolio passes 1,063 of 1,152 registered tests. Capacity checks and a separately reported post-hoc fitting analysis show why representability, finite-budget learning, calibration selection and validation coverage require distinct evidence.
Related research programmes
Cite this paper
Jeremy Rodgers. (2026). Learning Effective Interfaces from Opaque Stochastic Systems: Capacity, Selection, and Validation Limits (Version 2). https://doi.org/10.5281/zenodo.23075824