Skip to content
DraconiticResearch
Search Sources API
197,768 studies31 sourcesUpdated 26 min ago
PreprintarXiv licencearXiv

Basin Geometry and Reliable Recall of Dynamical Memories in Reservoir Computing

  • Ling-Wei Kong
  • Ying-Cheng Lai
Open at sourcePDFDOICite

Abstract

Reliable attractor recall conventionally requires broad basins of attraction. However, in reservoir-computing based associative memory, temporal cues reliably recover dynamical memories despite basins dominated by unpredictable, riddled-like regions. We reveal that memory basins exhibit an ``octopus-like'' structure: a robust ``head'' near the attractor and thin, intertwined ``tentacles'' spanning state space. Initial states in tentacular regions yield near-zero uncertainty exponents, making the recalled memory effectively unpredictable at finite precision. Yet, cue-driven generalized synchronization bypasses this unpredictability, driving the system into the robust basin head. This mechanism yields a quantitative relation linking minimum cue duration, synchronization rate, and basin-head radius. Trained recurrent neural networks exhibit similar geometry, suggesting this phenomenon extends beyond reservoir computing.

Related

  1. Inference of Unknown Dynamical Components Using Next Generation Reservoir Computing: From Chaotic Systems to Climate Data
    Jule Budnick, Andrew Keane, Serhiy Yanchuk
    arXivPreprintarXiv licenceFull textStatistical & nonlinear physicsarXiv
  2. Digital Twin Modeling of Quantum Dynamical Systems: Dissipative Quantum Reservoir Computing
    Abhijit Sen, Bikram Keshari Parida, Shital Chauhan +2
    arXivPreprintCC BY 4.0Full textQuantum physics & quantum informationarXiv
  3. Dynamical Diversity for Reservoir Computing in Reconfigurable Nanomechanics
    Humayun Ahmed, Inês S. Garcia, Filipa C. Mota +7
    arXivPreprintCC BY-NC-ND 4.0Full textApplied physics & instrumentationarXiv
  4. When Does Interference Help Learning? Kernel Geometry as a Pre-Experimental Test for Photonic Reservoir Computing
    Oishik Kar, Aswath Babu H
    arXivPreprintCC0Full textQuantum physics & quantum informationarXiv
  5. Reservoir computing generalized: Exploiting complex dynamics via time-invariant transformation
    Tomoyuki Kubota, Sumito Tsunegi, Kohei Nakajima
    Physical Review ResearchPeer-reviewedPublisher termsAtomic, molecular & optical physicsCrossref

Cite

BibTeXRIS
BibTeX
@misc{kong2026basin,
  title = {{Basin Geometry and Reliable Recall of Dynamical Memories in Reservoir Computing}},
  author = {Ling-Wei Kong and Ying-Cheng Lai},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.01914},
  eprint = {2609.01914},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.01914},
}
RIS
TY  - UNPB
TI  - Basin Geometry and Reliable Recall of Dynamical Memories in Reservoir Computing
AU  - Ling-Wei Kong
AU  - Ying-Cheng Lai
PY  - 2026
DA  - 2026/09/01
PB  - arXiv
DO  - 10.48550/arxiv.2609.01914
UR  - https://arxiv.org/abs/2609.01914
AB  - Reliable attractor recall conventionally requires broad basins of attraction. However, in reservoir-computing based associative memory, temporal cues reliably recover dynamical memories despite basins dominated by unpredictable, riddled-like regions. We reveal that memory basins exhibit an ``octopus-like'' structure: a robust ``head'' near the attractor and thin, intertwined ``tentacles'' spanning state space. Initial states in tentacular regions yield near-zero uncertainty exponents, making the recalled memory effectively unpredictable at finite precision. Yet, cue-driven generalized synchronization bypasses this unpredictability, driving the system into the robust basin head. This mechanism yields a quantitative relation linking minimum cue duration, synchronization rate, and basin-head radius. Trained recurrent neural networks exhibit similar geometry, suggesting this phenomenon extends beyond reservoir computing.
ER  -