Neural dynamics emulation using neuromorphic chips

Abstract: Recently a rigorous mathematical study of stochastic dynamics of the brain has shown that there exists a weak noise induced new phase, referred to as the noise-induced phase (N-phase) in between the other two phases: thermodynamic equilibrium (T-phase), and the ordinary chaos (C-phase). A healthy brain must reside in this N-phase, which collapses to a boundary, known previously as self-organized criticality. It is also shown that the N-phase can process information most effectively. In this paper, we used a neuromorphic chip “Spikey” to verify the presence of these three phases. We found good agreement between our emulation data and clinical data of the presence of the three phases. We further showed that the dynamics could be manipulated by controlling the avalanche parameters, such as the threshold of spiking. These results suggest that the healthy brain must operate in the generalized full N-phase rather than in the previously believed boundary or ‘edge of chaos’. Furthermore, we proposed that based on our preliminary work and results, such N-phase phase as in the normal brain is optimal for information processing and optimization problem. These results may pave the way for new computing paradigms using stochastic spiking dynamics of networks.

4819 Emperor Blvd, Suite 300 Durham, NC 27703 Voice: (919) 941-9400 Fax: (919) 941-9450