Attractor-merging Crises and Intermittency in Reservoir Computing
This addresses a fundamental dynamical systems problem in neural networks for researchers in computational neuroscience and machine learning, but it is incremental as it builds on known attractor embedding in reservoir computing.
The study identified an attractor-merging crisis with intermittency in reservoir computing networks by adjusting a global parameter, revealing its mechanism through phase-space analysis and showing it is intrinsic to a general class of random neural networks, independent of training data.
Reservoir computing can embed attractors into random neural networks (RNNs), generating a ``mirror'' of a target attractor because of its inherent symmetrical constraints. In these RNNs, we report that an attractor-merging crisis accompanied by intermittency emerges simply by adjusting the global parameter. We further reveal its underlying mechanism through a detailed analysis of the phase-space structure and demonstrate that this bifurcation scenario is intrinsic to a general class of RNNs, independent of training data.