Reservoir Computing - Machine Learning and Dynamical Systems

Explore cutting-edge developments in reservoir computing through this comprehensive 8-hour symposium featuring twelve specialized presentations on machine learning and dynamical systems. Delve into diverse implementations of reservoir computing including autonomous Boolean networks on FPGAs, superconducting circuits, and photonic systems with time and wavelength multiplexing. Examine theoretical foundations through embedding and approximation theorems for echo state networks, network statistics analysis, and explanations of reservoir computing's surprising forecasting success in chaotic systems. Investigate practical applications spanning turbulent and geophysical flow prediction with scale separation techniques, financial modeling using randomized signatures, and high-speed hardware acceleration methods. Learn about advanced concepts including multistability in input-driven recurrent neural networks, synchronization mechanisms for measuring echo state properties, and the mathematical underpinnings of reservoir computing through randomized discrete-time signatures.

Syllabus

Reservoir Computing with Autonomous Boolean Networks on Field Programmable Gate Arrays
Reservoir Computing with Superconducting Circuits
Boosting performance in Machine Learning of Turbulent and Geophysical Flows via scale separation
On Explaining the Surprising Success of Reservoir Computing Forecaster of Chaos?
Reservoir computing: prediction and high-speed hardware accelerators
Network Statistics for Reservoir Computing
Learn to Synchronize, Synchronize to Learn: measuring the Echo State Property
Multistability in input-driven recurrent neural networks
Embedding and Approximation Theorems for Echo State Networks
Explaining the reservoir computing phenomenon using randomized discrete-time signatures
Randomized Signature and Reservoir Computing with application to Finance
Time- and Wavelength-Multiplexed Photonic Reservoir Computing