analiza aiqbench
This research introduces Dynamical Mode Pruning (DMP), a novel technique for optimizing Echo State Networks by evaluating neurons based on their contribution to input-driven state transitions rather than static metrics. By leveraging trajectory-averaged Jacobian Gramians, the method effectively identifies and removes redundant components in reservoir computing models without sacrificing forecasting accuracy on chaotic time-series benchmarks. The study highlights that dynamic influence is a critical criterion for refining over-parameterized neural architectures beyond traditional structural importance measures.
Rethinking Reservoir Pruning: A Dynamical Perspective for Echo State Networks
arXiv:2608.04593v1 Announce Type: new Abstract: Echo State Networks (ESNs) offer an efficient framework for temporal prediction, but their randomly initialized reservoirs are often over-parameterized and dynamically redundant. Existing pruning methods largely rely on static connectivity or activation statistics, which may overlook neurons that shape input-driven state transitions. We propose Dynamical Mode Pruning (DMP), a reservoir pruning method that ranks neurons by their contribution to dominant transition modes obtained from a trajectory-averaged Jacobian Gramian. DMP removes low-impact units and retrains only the readout. Experiments on chaotic and real-world time-series benchmarks show that DMP improves or preserves forecasting accuracy while reducing redundant reservoir components. Our results suggest that dynamical influence is a useful criterion for reservoir refinement beyond static structural importance alone.
Abstract:Echo State Networks (ESNs) offer an efficient framework for temporal prediction, but their randomly initialized reservoirs are often over-parameterized and dynamically redundant. Existing pruning methods largely rely on static connectivity or activation statistics, which may overlook neurons that shape input-driven state transitions. We propose Dynamical Mode Pruning (DMP), a reservoir pruning method that ranks neurons by their contribution to dominant transition modes obtained from a trajectory-averaged Jacobian Gramian. DMP removes low-impact units and retrains only the readout. Experiments on chaotic and real-world time-series benchmarks show that DMP improves or preserves forecasting accuracy while reducing redundant reservoir components. Our results suggest that dynamical influence is a useful criterion for reservoir refinement beyond static structural importance alone.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Dynamical Systems (math.DS)
arXiv:2608.04593 [cs.LG]
(or arXiv:2608.04593v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.04593 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 215 cuvinte
Subiecte conexe
Echo Flow Networks
arXiv · 6 aug. 2026
Analyzing the Dynamics of WSBM-based Echo State Networks using InputDSA
arXiv · 15 oct. 2026
Fractal and Chaotic Activation Functions in Echo State Networks: Preprocessing Topology Governs the Echo State Property
arXiv · 7 sept. 2026
Lindblad-Inspired Multi-Timescale Reservoir Computing with Separable Rotation and Dissipation
arXiv · 6 aug. 2026