Montejo N, Lorenzo M N, Pérez-Villar V, Pérez-Muñuzuri V
Group of Nonlinear Physics, Faculty of Physics, University of Santiago de Compostela, Spain.
Phys Rev E Stat Nonlin Soft Matter Phys. 2005 Jul;72(1 Pt 1):011902. doi: 10.1103/PhysRevE.72.011902. Epub 2005 Jul 5.
The role of spatially correlated stochastic perturbations on a Morris-Lecar neural network subject to an aperiodic subthreshold signal is analyzed in detail. Our results suggest that optimum signal-to-noise ratios can be obtained for two critical noise intensities due to the interplay of the subthreshold Poisson process and the correlated Gaussian forcing. For the second peak, most of the cells are periodically excited, the information transfer is enhanced, and a collective behavior develops measured in terms of the averaged activity of the network. The maximum signal-to-noise ratio increases with the correlation length, although it saturates for global coupling. It was found that there is a range of mean frequencies of the subthreshold signal that increases the signal-to-noise ratio output.
详细分析了空间相关随机扰动对受非周期阈下信号作用的莫里斯 - 莱卡神经网络的作用。我们的结果表明，由于阈下泊松过程与相关高斯强迫的相互作用，对于两个临界噪声强度可获得最佳信噪比。对于第二个峰值，大多数细胞被周期性激发，信息传递增强，并且根据网络的平均活动来衡量会出现集体行为。最大信噪比随相关长度增加，尽管对于全局耦合它会饱和。研究发现，存在一定范围的阈下信号平均频率会增加输出信噪比。
