回声状态网络

A type of recurrent neural network (RNN) that relies on a fixed, randomly initialized dynamic system known as the reservoir to transform input signals into a high-dimensional state space. The reservoir's dynamics are frozen, and only the readout weights are trained to map these states to the desired output. / 一种依赖于固定、随机初始化的动态系统（即回声状态）将输入信号转换为高维状态空间的循环神经网络（RNN）。回声状态动力学被冻结，仅训练读出权重以将这些状态映射到所需的输出。

Knowledge type

数学理论

First proposed or discovered

Proposer

Turkey Farm member

Creator

农场主

Last updated

2026-11-06 23:06:27