Teaching neural networks to process temporal signals for oculomotor control.
Teaching neural networks to process temporal signals for oculomotor control.
复制标题
教授神经网络处理用于动眼神经控制的时间信号。
DOI：
10.1101/sqb.1990.055.01.077
复制
发表时间：
1990
期刊：
影响因子：
--
通讯作者：
Robinson,DA
中科院分区：
文献类型：
--
作者：
Arnold,DB;Robinson,DA
方法最近研究的许多神经网络（Rumelhart et al. 1986）被设计用于学习识别空间输入模式。它们不接受或产生随时间变化的信号。为了获得这样的信号，必须用一阶微分方程描述每个神经元，然后对n个神经元求解n个联立方程。每个神经元对输入之和的线性响应与它们的突触权重成比例。
METHODSMany of the neural networks that have been studied recently (Rumelhart et al. 1986) are designed to learn to recognize spatial input patterns. They do not accept or produce signals that are functions of time. To obtain such signals, one must describe each neuron by a firstorder differential equation and then, for n neurons, solve n simultaneous equations. Each neuron responded linearly to the sum of its inputs in proportion to their synaptic weights.