MATHEMATICAL-ANALYSIS OF A NEURAL NETWORK WITH INHIBITORY COUPLING

STOCHASTIC PROCESSES AND THEIR APPLICATIONS

卷:40

Article

COTTRELL, M

关键词: MARKOV CHAINS;    NEURAL NETWORKS;    INHIBITORY COUPLING;    CEREBELLAR CORTEX;    STIMULI;   

DOI  :  10.1016/0304-4149(92)90140-L

来源: Elsevier

【 摘 要 】

We study the role of inhibition in a nearest-neighbours-connected neural model. The state of the network is a Markov process of which we study the ergodic properties or divergence characteristics using the parameters of the system. We prove that, when inhibition is smaller than a certain threshold, the network is ergodic and works in a stationary way. Conversely, when inhibition increases, the network is divided into two groups: active and inactive neurons. We observe by means of computer simulation that striped or moire responses appear, whose shape and width depend on considered neighbourhood size. The model resembles the biological reality of the young animal's cerebellar cortex.

【 授权许可】

Free