Study Of Brain Network Dynamics Based On A Chaotic Neural Network

A human brain is the most sophisticated system in the world. A huge number of neurons, neuron clusters and brain areas in this system connect each other to form a complex network which realizes various functions of brain by synergy effect. Now scientists are using modern electromagnetic technology, brain imaging technology, statistical physics theory and complex network theory to build brain networks, and study their topology structures. It is revealed that the topology structure of a brain network is related to its cognitive levels and brain diseases of the tester. But there is no report about neural dynamics of a brain network so far.An artificial neural network is a description of human brain information processing characteristics. It has functions of associative memory, non-linear mapping, classification recognition, optimization calculation and so on. Animal neural electrophysiological experiments have shown that chaos exists in biological neural systems,Aihara chaotic neural network reflects the chaotic characteristics of nervous systems and has been widely concerned.Chaotic behaviors exist in brain neural systems.In this paper, we introduce Aihara CNN to a functional brain network constructed by fMRI brain imaging techniques and study its dynamics behaviors. We regard the topology structure of the brain network as that of a CNN,brain areas as neural CNN's nodes,correlation coefficients of brain areas as the connection weights between nodes.We found that brain areas show different dynamic behaviors and can be roughly divided into three categories,active brain areas whose outputs change rapidly between 0 and 1 rapidly,dumb brain areas whose outputs are stable in a certain value and less active brain whose outputs change slowly. Besides,the corresponding brain areas for left and right brain have different dynamics behaviors,which suggest different functions of left and right brain.In the construction of functional brain network, threshold value of correlation coefficients impact on the network topology structure. It is controverted about how to select threshold values. We investigated effects of threshold values on network dynamics. It is revealed that there is little influence on the dynamic characteristics of brain network when the threshold value of connection weights is less than 0.2. We studied the dynamics characteristics of brain network with different Aihara CNN coefficients, and observed that there are two different dynamics phases in the brain network, chaos and order. The output sequence patterns in brain networks also confirm this result.