Reservoir Computing
Overview
Reservoir computing studies how the evolving states of recurrent systems can encode useful information about time-varying inputs, often without training the recurrent connections. Common approaches include echo state networks with fitted linear readouts, next-generation reservoirs built from delays and nonlinear features, and physical reservoirs based on devices such as photonic, memristive, mechanical, or biological systems. Researchers apply these methods to chaotic-system forecasting, signal classification, control, and scientific modeling, while improving them through changes to topology, timescales, pruning, and online adaptation. A central challenge is balancing memory, stability, and expressive dynamics; for quantum reservoirs in particular, broad advantages over carefully matched classical methods remain unestablished.
Latest papers
114 papers12 papers in the last four weeks, up 20% on the four weeks before. 0.1% of all new papers.
Anomaly Detection on Resource-Constrained Devices with Reservoir Computing
Is Newborn Wiring Special? Separating Wiring from Weight Placement in Connectome Reservoirs
RC Prediction: A Python Package for Parameter-Aware Reservoir Computing
MulClip: A table-free activation function with monotonic and folding regimes for reservoir computing on microcontrollers
Beyond Low-Pass Dynamics: Frequency-Selective Spiking Reservoirs with Resonant Neurons
Extrapolating learned dynamics across bifurcations based on generalized synchronization
Optimal Connectivity of Random Boolean Networks for Reservoir Computing
Sector-Ring Linear Reservoirs: Sector-Wise Decay-Timescale Allocation With State Reuse Across Candidate Settings
CGDT-CESN: Correlation-Guided and Dynamics-Trained Clustered Echo State Network for Multivariate Time-Series Prediction
Empirical Boundaries of Reservoir Computing for Few-Shot Sequence Pattern Tasks
Collective Benefits and Efficiency in Coupled Echo State Networks
Analyzing the Dynamics of WSBM-based Echo State Networks using InputDSA
Does Learning Recurrent Networks Matter in Partially Observable RL?
Matching of signal, noise and hardware timescales for filtering and forecasting of correlated noise signals
A KKL Observer Perspective on Reservoir Computing
Context-dependent time-series prediction via HyperReservoirs
Uncertainty Quantification of Next Generation Reservoir Computing with Applications to Memory-Driven Dynamical Systems
Benchmarking the Connectomes of Caenorhabditis elegans within the Reservoir Computing Framework
Dynamical Diversity for Reservoir Computing in Reconfigurable Nanomechanics
Inference of Unknown Dynamical Components Using Next Generation Reservoir Computing: From Chaotic Systems to Climate Data
Shake to Learn: Dynamic Interrogation of Hidden Object Physics for Robotic Manipulation with Physical Reservoir Computing
LoRA-RC: Reservoir Computing with Low-Rank Adaptation
Beyond density snapshots: Dynamical memory in Chern-Simons reservoirs
Photonic reservoir computing with dimensionally compressed readout
Connectome-to-Function: Conditional Generative Latent Representations for Reservoir Computing
Towards Effective Physical Reservoir Computing with a Pneumatic Soft Robot
Basin Geometry and Reliable Recall of Dynamical Memories in Reservoir Computing
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.