The Research Of The Simulation Of The Visual Cortex Networks Evolution

Visual system is one of the most complex and efficient system. Research found that visual system to perform any function is completed by the mutual connections coordination and interaction between a number of neurons or more clusters or neurons in the visual cortex. Proposed for complex network provides a good basis for the study of complex system, so that from the perspective of the network to study the dynamics of the structural characteristics of the visual system, for our understanding of the formation mechanism and its internal research on visual intelligence technology is of great significance.Visual system is a high-energy systems, the work of the visual cortex network should try to reduce this consumption that is the cost of wiring. However, the visual cortex network has higher topological properties, robustness, modular and large connecting hub to complete the high efficiency of information integration. Therefore, the breakthrough point is how to balance the minimum wiring cost and the global optimal efficiency in this paper, first of all, we define the edge probability function according to the space distance based on strict space constraints, use two dimensional homogeneity growth model to simulate the real visual system, and We analysis the visual cortex network topology characteristics. Second, we carried out on the network evolution according to the distribution of heterogeneous neurons and preferential attachment mechanism, and explore the changes of wiring cost and network topology with the change of the scale factor. Finally, making use of complex network game method we carry out on the network game based on the neuron’s limited memory, and analyze the influence which memory length made to collaborators density and network topology. This article reveals the inner core mechanism of visual information processing from the network point of view, changing the previous studies of visual information processing from the fundamental ideas and methods.(1) From the visual network is a typical space network to start, according to the multi-dimensional time and space scales, we give the connection probability function to simulate real network model of the visual cortex. The local and overall network topology characteristics were analyzed to verify the visual cortex with a small-world network.(2) On the basis of the above conclusions, we study the impact on visual cortical network topology of the uneven distribution of the number of heterogeneity neurons. Then on the basis of the heterogeneity neurons to introduce preferential attachment mechanism, namely the synaptic priority connection and long distance priority connection, we analyze its impact on maximizing small world characteristics in the time scale and spatial scale.(3) On the basis of neuronal limited memory and the balance between minimum wiring cost and the global optimal efficiency, we have set a new function by using evolutionary game theory, we have the network game using the most update rules. We study the cooperation density changes by changing the length of memory, and show network topology of the stable performance after the evolutionary game with network visualization, and analyze the interaction between its network features and game dynamics based on the time scales and space scales.