Hong Kong Baptist University

2) Self-organized activity and information processing in complex neural networks

Cerebral cortical brain networks possess a number of prominent features of structure and dynamics. Firstly, they are structured as hierarchical modular networks, from large-scale regions of the brain, via cortical areas and area sub-compartments organized as structural and functional maps, to cortical columns, and finally circuits made up of individual neurons. A balance of excitation and inhibition also features local recurrent neural networks. Second, the networks display self-organized sustained activity, which is persistent in the absence of external stimuli. At systems level, such activities are characterized by complex rhythmical oscillations over a broadband background, while at the cellular level or mesoscales, neuronal discharges display avalanches, indicating that cortical networks are at the self-organized critical states. We are interested in understanding how such highly fluctuating activities are organized in biologically realistic complex neural network models and how they are employed in neural information processing and learning. As the first step, we have shown that hierarchical modular network organization has important impact on self-organized critical dynamics. Our research will be focusing on the impact of such fluctuating spatial-temporal patterns on the timing of neuronal firing, and on the structure-dynamics co-evolution due to neural plasticity, different from information processing using attractor dynamics in connectionist's artificial neural network models.

Related publications:
S.J. Wang and C.S. Zhou, "Hierarchical modular structure enhances the robustness of self-organized criticality in neural networks" , New J. Phys. 14, 023005 (2012).
S. Wang, C.C. Hilgetag and C.S. Zhou, "Sustained activity in hierarchical modular neural networks: self-organized criticality and oscillations", Frontiers in Computational Neuroscience, 5, 30 (2011)

Back to Research Webpage