Noise-Driven Temporal Association in Neural Networks
Noise-Driven Temporal Association in Neural Networks
复制标题
神经网络中噪声驱动的时间关联
DOI：
--
复制
发表时间：
1987
期刊：
影响因子：
--
通讯作者：
K. Schulten
中科院分区：
文献类型：
--
作者：
J. Buhmann;K. Schulten
提出了一种具有非对称交换作用和随机锋电位反应的类自旋神经元网络，它可以学习和回忆有偏模式的时间序列。噪声使得具有延迟响应或具有时间依赖强度的突触变得多余，这些突触先前被提议用于存储时间序列。模式序列的精确定时需要足够数量的N个神经元。该网络的性能通过Monte Carlo模拟来描述，根据Fokker-Planck方程，当N → ∞时，根据Liouville方程。
A network of spinlike neurons with asymmetric exchange interactions and stochastic spike response which can learn and recall time sequences of biased patterns is proposed. Noise makes synapses with delayed response or with time-dependent strength, previously proposed for storage of time sequences, superfluous. An accurate timing of pattern sequences requires a sufficient number N of neurons. The performance of the suggested network is described by Monte Carlo simulation, in terms of a Fokker-Planck equation and, for N → ∞, in terms of a Liouville equation.