Diagnostic Benchmarks for Invariant Learning Dynamics: Empirical Validation of the Eidos Architecture

TL;DR

This paper introduces the PSI dataset to evaluate topological invariance in models, demonstrating that the Eidos architecture maintains structural identity across transformations with high accuracy, supporting the 'Form-First' hypothesis.

Contribution

The paper presents the PSI benchmark suite and empirically validates the Eidos architecture's ability to achieve invariance through geometric integrity rather than statistical scale.

Findings

Eidos achieves >99% accuracy on PSI benchmarks.

Eidos attains 81.67% zero-shot transfer on unseen typefaces.

Results support the 'Form-First' hypothesis linking invariance to geometric integrity.

Abstract

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.