Embedding theory of reservoir computing and reducing reservoir network using time delays
Xing-Yue Duan, Ying Xiong, Siyang Leng, Jürgen Kurths, Wei Lin, Huanfei Ma
Abstract
The dynamics of reservoir computing (RC), a compact recurrent neural network, is validated as a higher-dimensional embedding of the input nonlinear dynamics. Based on this rigorous validation, a delayed RC is established with a significantly reduced network size and a promoted memory capacity, which can achieve dynamics reconstruction even in the reservoir with a single neuron.
Topics & Concepts
Reservoir computingEmbeddingComputer scienceNonlinear systemReservoir modelingDynamics (music)Reservoir simulationArtificial neural networkRecurrent neural networkDistributed computingPetroleum engineeringParallel computingArtificial intelligenceGeologyPhysicsAcousticsQuantum mechanicsNeural Networks and Reservoir ComputingAdvanced Memory and Neural ComputingNeural dynamics and brain function