Blind Dynamical Taxonomy: A Pre-Registered Failure Limits of Feature-Based Classification Across Canonical Dynamical Systems

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Author: Mitchell , Thomas S.
Format: Recurso digital
Language:English
Published: Zenodo 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866901370922074112
author Mitchell , Thomas S.
author_facet Mitchell , Thomas S.
contents <p>Pre-registered negative-results study testing whether canonical dynamical systems can be classified from observable time-series features alone. Twenty-one systems spanning seven dynamical families were embedded using 22 extracted features across spectral, temporal, recurrence-quantification, transient-response, morphology, and embedding domains.<br>The experiment failed its pre-registered success threshold. The strongest failure occurred within the Gray-Scott reaction-diffusion family itself, where systems generated by the same governing PDE failed to cluster together under parameter variation.<br>Results demonstrate that parameter variation within a fixed mechanism family can overwhelm observable feature similarity in low-dimensional embedding space. A companion naive-feature experiment produced complete collapse into a single dominant cluster.<br>The paper discusses why feature-based dynamical taxonomy fails under projection loss, observable dependence, and parameter sensitivity, and outlines what types of invariant representations may be required for robust classification.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_20084204
institution Zenodo
language eng
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle Blind Dynamical Taxonomy: A Pre-Registered Failure Limits of Feature-Based Classification Across Canonical Dynamical Systems
Mitchell , Thomas S.
dynamical systems, time-series analysis, feature extraction, clustering, reaction-diffusion systems, Gray-Scott model, Kuramoto model, Ising model, recurrence quantification analysis, attractor reconstruction, temporal geometry, Keller-ΔΦ, nonlinear dynamics, complexity science, observability, embedding space, phase transitions, topology, dynamical taxonomy, negative results
<p>Pre-registered negative-results study testing whether canonical dynamical systems can be classified from observable time-series features alone. Twenty-one systems spanning seven dynamical families were embedded using 22 extracted features across spectral, temporal, recurrence-quantification, transient-response, morphology, and embedding domains.<br>The experiment failed its pre-registered success threshold. The strongest failure occurred within the Gray-Scott reaction-diffusion family itself, where systems generated by the same governing PDE failed to cluster together under parameter variation.<br>Results demonstrate that parameter variation within a fixed mechanism family can overwhelm observable feature similarity in low-dimensional embedding space. A companion naive-feature experiment produced complete collapse into a single dominant cluster.<br>The paper discusses why feature-based dynamical taxonomy fails under projection loss, observable dependence, and parameter sensitivity, and outlines what types of invariant representations may be required for robust classification.</p>
title Blind Dynamical Taxonomy: A Pre-Registered Failure Limits of Feature-Based Classification Across Canonical Dynamical Systems
topic dynamical systems, time-series analysis, feature extraction, clustering, reaction-diffusion systems, Gray-Scott model, Kuramoto model, Ising model, recurrence quantification analysis, attractor reconstruction, temporal geometry, Keller-ΔΦ, nonlinear dynamics, complexity science, observability, embedding space, phase transitions, topology, dynamical taxonomy, negative results
url https://doi.org/10.5281/zenodo.20084204