We present the PolyShapes-Ideal (PSI) dataset, a suite of diagnostic benchmarks designed to isolate topological invariance -- the ability to maintain structural identity across affine transformations -- from the textural correlations that dominate standard vision benchmarks. Through three diagnostic probes (polygon classification under noise, zero-shot font transfer from MNIST, and geometric collapse mapping under progressive deformation), we demonstrate that the Eidos architecture achieves >99% accuracy on PSI and 81.67% zero-shot transfer across 30 unseen typefaces without pre-training. These results validate the "Form-First" hypothesis: generalization in structurally constrained architectures is a property of geometric integrity, not statistical scale.
Cite
@article{arxiv.2602.13322,
title = {Diagnostic Benchmarks for Invariant Learning Dynamics: Empirical Validation of the Eidos Architecture},
author = {Datorien L. Anderson},
journal= {arXiv preprint arXiv:2602.13322},
year = {2026}
}
Comments
8 pages, 3 figures and extra material to help can be found: https://zenodo.org/records/18529180