arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1710.04400q-bio.NC
收藏

通过收敛至复杂振荡放电模式实现的回响神经网络信息约简

Information reduction in a reverberatory neuronal network through convergence to complex oscillatory firing patterns

A. Vidybida, O. Shchur

AI总结:

本研究通过模拟9个神经元组成的全连接回响神经网络,发现83%的初始刺激会使网络收敛至共102种不同的周期放电终态,神经元放电是轨迹合并的必要条件,该自组织现象可为感知与记忆机制提供模型参考。

摘要:

我们通过计算机模拟研究了回响神经网络的动力学特性。该网络由9个以正方形晶格排列的漏泄积分放电(LIF,leaky integrate-and-fire)神经元组成,为全连接结构,神经元间通信延迟与对应距离成正比。网络先接受不同刺激的初始刺激,随后进入自由演化状态。对于每个刺激,在自由演化过程中,神经活动要么完全消失,要么网络收敛至一条周期轨迹,不同刺激对应的周期轨迹可能不同。在包含285290个初始刺激的集合中观察到了后者(即收敛至周期轨迹的情况),该集合占所有施加刺激的83%。通过施加该集合中的每个刺激,我们共发现了102种不同的周期终态。通过分析轨迹我们得出结论:神经元放电是不同轨迹合并为单条轨迹的必要前提,合并后的轨迹最终会转变为周期状态。我们讨论了观察到的时域自组织现象,认为其可作为感知过程中发生的相关过程的潜在模型;周期状态下的重复放电可能是记忆形成的基础。

英文摘要:

We study dynamics of a reverberating neural net by means of computer simulation. The net, which is composed of 9 leaky integrate-and-fire (LIF) neurons arranged in a square lattice, is fully connected with interneuronal communication delay proportional to the corresponding distance. The network is initially stimulated with different stimuli and then goes freely. For each stimulus, in the course of free evolution, activity either dies out completely or the network converges to a periodic trajectory, which may be different for different stimuli. The latter is observed for a set of 285290 initial stimuli which constitutes 83% of all stimuli applied. By applying each stimulus from the set, we found 102 different periodic end-states. By analyzing the trajectories, we conclude that neuronal firing is the necessary prerequisite for merging different trajectories into a single one, which eventually transforms into a periodic regime. Observed phenomena of self-organization in the time domain are discussed as a possible model for processes taking place during perception. The repetitive firing in the periodic regimes could underpin memory formation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑