EUVIMEDEUVIMED entdecken
Crossref

Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing

Claus Metzner, Achim Schilling, Andreas Maier, Patrick Krauss

Neural Computation · 17.07.2025 · 37 (8) · 1469-1504

DOI: 10.1162/neco_a_01770 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Neural Computation, online erschienen am 17.07.2025. Verfasst von Claus Metzner und 3 weiteren Autor:innen. Beteiligte Einrichtung: Pattern-Recognition Lab, Friedrich-Alexander-University, 91058 Erlangen, Germany claus.metzner@fau.de. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 47 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Reservoir computing information processing based on untrained recurrent neural networks with random connections is expected to depend on the nonlinear properties of the neurons and the resulting oscillatory, chaotic, or fixed-point dynamics of the network. However, the degree of nonlinearity required and the range of suitable dynamical regimes for a given task remain poorly understood. To clarify these issues, we study the classification accuracy of a reservoir computer in artificial tasks of varying complexity while tuning both the neuron’s degree of nonlinearity and the reservoir’s dynamical regime. We find that even with activation functions of extremely reduced nonlinearity, weak recurrent interactions, and small input signals, the reservoir can compute useful representations. These representations, detectable only in higher-order principal components, make complex classification tasks linearly separable for the readout layer. Increasing the recurrent coupling leads to spontaneous dynamical behavior. Nevertheless, some input-related computations can “ride on top” of oscillatory or fixed-point attractors with little loss of accuracy, whereas chaotic dynamics often reduces task performance. By tuning the system through the full range of dynamical phases, we observe in several classification tasks that accuracy peaks at both the oscillatory/chaotic and chaotic/fixed-point phase boundaries, supporting the edge of chaos hypothesis. We also present a regression task with the opposite behavior. Our findings, particularly the robust weakly nonlinear operating regime, may offer new perspectives for both technical and biological neural networks with random connectivity.

Autor:innen und Institutionen

  • Pattern-Recognition Lab, Friedrich-Alexander-University, 91058 Erlangen, Germany claus.metzner@fau.de.
  • Pattern-Recognition Lab, Friedrich-Alexander-University, 91058 Erlangen, Germany achim.schilling@fau.de.
  • Pattern-Recognition Lab, Friedrich-Alexander-University, 91058 Erlangen, Germany andreas.maier@fau.de.
  • Pattern-Recognition Lab, Friedrich-Alexander-University, 91058 Erlangen, Germany patrick.krauss@fau.de.

Zitieren

APA

Metzner, C., Schilling, A., Maier, A., & Krauss, P. (2025). Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing. Neural Computation, 37(8), 1469-1504. https://doi.org/10.1162/neco_a_01770

Vancouver

Metzner C, Schilling A, Maier A, Krauss P. Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing. Neural Computation. 2025;37(8):1469-1504. doi:10.1162/neco_a_01770

BibTeX
@article{Metzner2025,
  author = {Metzner, Claus and Schilling, Achim and Maier, Andreas and Krauss, Patrick},
  title = {Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing},
  journal = {Neural Computation},
  year = {2025},
  volume = {37},
  number = {8},
  pages = {1469--1504},
  doi = {10.1162/neco_a_01770}
}
Kostenlos starten →