Answers are grounded in this paper and cite their sources.
This preprint establishes a compression regime in neural networks influencing chaotic systems, suggesting significant implications for data reduction.
This preprint establishes an empirical scaling law for neural network compression of strange attractors. The primary validated result is the compression regime: D* ≈ 0.5–0.7 × d_KY across 27 systems satisfying Strange Attractor Primacy conditions, representing 30–50% compression beyond linear methods (POD). The discovery regime formula D* = α·log(n/m) + β is theoretically grounded in Shannon's rate-distortion theory, which uniquely determines the logarithmic form. However, direct measurement of α is challenging for low-dimensional systems, and specific α values (theory predicts ~48 for ODEs, ~54 for maps) await rigorous experimental protocols. Training dynamics independently validate category structure (ANOVA p=0.0008) and SAP boundary conditions (12× convergence speedup for z-kurtosis > 5, p < 10⁻⁵). A full theoretical derivation of the ~0.5 compression factor from information-geometric principles is ongoing work.
No takes yet. Share an insight, caveat, or question.
Jon Wiberg (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: