基于突触离子通道动力学神经元网络的高效并行仿真算法
|
摘要
在计算神经科学领域,大规模神经元网络的并行仿真对探索和揭示生物大脑中信息传递机制有着重要作用。为加速大规模神经元网络仿真,提出一种模块独立性强、耦合度低的基于突触递质-受体离子通道动力学的神经元网络的并行算法。通过分析化学突触信息传递机理及递质分子、受体离子通道动力学特征,提出了递质-受体计算分离的思想,增强了突触前神经元引起的递质分子浓度计算与突触后绑定状态的受体浓度计算之间的独立性,降低突触电流计算中突触前神经元状态和突触后神经元状态之间的耦合度。基于上述思想,设计并实现了一种生物神经网络并行算法。仿真结果表明了该算法的高效性。
In the field of computational neuroscience,the parallel simulation of large-scale neuronal networks plays a very important role in exploring and revealing the information processing mechanism of the brain.In order to accelerate the large-scale neuronal network simulation,an efficient parallel neuronal network algorithm based on synaptic neurotransmitter-receptor ion channel kinetic characteristics is proposed.By analyzing the information transmission mechanisms and the neurotransmitter-receptor ion channel kinetic characteristics of the chemical synapse,we propose an idea of the separation of presynaptic neurotransmitter computing from postsynaptic receptor computing.The new idea enhances the independence of the two concentration calculation:neurotransmitter emitted by presynaptic neurons,and the bound postsynaptic receptors.During each synaptic current computing,the new idea reduces the coupling degree of the presynaptic neuron and the postsynaptic neuron.Based on the new idea,we design a parallel algorithm for parallel neuronal network simulation based on synaptic ion channel kinetic.Simulation results show that the algorithm is efficient.
In the field of computational neuroscience,the parallel simulation of large-scale neuronal networks plays a very important role in exploring and revealing the information processing mechanism of the brain.In order to accelerate the large-scale neuronal network simulation,an efficient parallel neuronal network algorithm based on synaptic neurotransmitter-receptor ion channel kinetic characteristics is proposed.By analyzing the information transmission mechanisms and the neurotransmitter-receptor ion channel kinetic characteristics of the chemical synapse,we propose an idea of the separation of presynaptic neurotransmitter computing from postsynaptic receptor computing.The new idea enhances the independence of the two concentration calculation:neurotransmitter emitted by presynaptic neurons,and the bound postsynaptic receptors.During each synaptic current computing,the new idea reduces the coupling degree of the presynaptic neuron and the postsynaptic neuron.Based on the new idea,we design a parallel algorithm for parallel neuronal network simulation based on synaptic ion channel kinetic.Simulation results show that the algorithm is efficient.
引文
[1]Basu I,Kudela P,Korzeniewska A,et al.A study of the dynamics of seizure propagation across micro domains in the vicinity of the seizure onset zone[J].Journal of Neural Engineering,2015,12(4):046016.
[2]Carnevale N T,Hines M L.The NEURON book[M].Cambridge:Cambridge University Press,2006.
[3]Kim D,ParéD,Nair S S.Mechanisms contributing to the induction and storage of Pavlovian fear memories in the lateral amygdala[J].Learning&Memory,2013,20(8):421-430.
[4]Jan M,Tomohiro S,Kenji D.The mechanism of saccade motor pattern generation investigated by a large-scale spiking neuron model of the superior colliculus[J].Plos One,2013,8(2):e57134.
[5]Markram H.The blue brain project[J].Nature Rev Neurosci,2006,7(2):153-160.
[6]Brette R,Rudolph M,Carnevale T,et al.Simulation of networks of spiking neurons:A review of tools and strategies[J].Journal of Computational Neuroscience,2007,23(3):349-398.
[7]Milkowski M.Explanatory completeness and idealization in large brain simulations:A mechanistic perspective[J].Synthese,2016,193(5):1457-1478.
[8]Chapeau-bloudeau F,Chambet N.Synapse models for neural networks:From ion channel kinetics to multiplicative coefficient wij[J].Nueral Computation,1995,7(4):713-734.
[9]Destexhe A,Mainen Z F,Sejnowski T J.Synthesis of models for excitable membranes,synaptic transmission and neuromodulation using a common kinetic formalism[J].Journal of Computational Neuroscience,1994,1(3):195-230.
[10]Hanuschkin A,Kunkel S,Helias M,et al.A general and efficient method for incorporating precise spike times in globally time-driven simulations[J].Frontiers in Neuroinformatics,2010,4(8):113.
[11]Lin Xiang-hong.Simulation and evolution of large-scale spiking neural networks[D].Harbin:Harbin Institute of Technology,2009.(in Chinese)
[12]Nageswaran J M,Dutt N,Krichmar J L,et al.2009special issue:A configurable simulation environment for the efficient simulation of large-scale spiking neural networks on graphics processors[J].Neural Networks the Official Journal of the International Neural Network Society,2009,22(5-6):791-800.
[13]Brette R,Goodman D F.Vectorized algorithms for spiking neural network simulation[J].Neural Computation,2011,23(6):1503-1535.
[14]Dayan P,Abbott L F.Theoretical neuroscience[M].Massachusetts:MIT Press,2001:154-155.
[15]Brette R.Exact simulation of integrate-and-fire models with synaptic conductances[J].Neural Computation,2006,18(8):2004-2027.
[16]Hodgkin A L,Huxley A F.Currents carried by sodium and potassium ions through the membrane of the giant axon of Loligo[J].Journal of Physiology,1952,116(4):449.
[17]Li Lian-jiang.Research on GPU based spiking neural network learning[D].Wuhan:Huazhong University of Science and Technology,2015.(in Chinese)
[11]蔺想红.大规模脉冲神经网络的模拟与进化研究[D].哈尔滨:哈尔滨工业大学,2009.
[17]李连江.基于GPU的脉冲神经网络学习研究[D].武汉:华中科技大学,2015.
[2]Carnevale N T,Hines M L.The NEURON book[M].Cambridge:Cambridge University Press,2006.
[3]Kim D,ParéD,Nair S S.Mechanisms contributing to the induction and storage of Pavlovian fear memories in the lateral amygdala[J].Learning&Memory,2013,20(8):421-430.
[4]Jan M,Tomohiro S,Kenji D.The mechanism of saccade motor pattern generation investigated by a large-scale spiking neuron model of the superior colliculus[J].Plos One,2013,8(2):e57134.
[5]Markram H.The blue brain project[J].Nature Rev Neurosci,2006,7(2):153-160.
[6]Brette R,Rudolph M,Carnevale T,et al.Simulation of networks of spiking neurons:A review of tools and strategies[J].Journal of Computational Neuroscience,2007,23(3):349-398.
[7]Milkowski M.Explanatory completeness and idealization in large brain simulations:A mechanistic perspective[J].Synthese,2016,193(5):1457-1478.
[8]Chapeau-bloudeau F,Chambet N.Synapse models for neural networks:From ion channel kinetics to multiplicative coefficient wij[J].Nueral Computation,1995,7(4):713-734.
[9]Destexhe A,Mainen Z F,Sejnowski T J.Synthesis of models for excitable membranes,synaptic transmission and neuromodulation using a common kinetic formalism[J].Journal of Computational Neuroscience,1994,1(3):195-230.
[10]Hanuschkin A,Kunkel S,Helias M,et al.A general and efficient method for incorporating precise spike times in globally time-driven simulations[J].Frontiers in Neuroinformatics,2010,4(8):113.
[11]Lin Xiang-hong.Simulation and evolution of large-scale spiking neural networks[D].Harbin:Harbin Institute of Technology,2009.(in Chinese)
[12]Nageswaran J M,Dutt N,Krichmar J L,et al.2009special issue:A configurable simulation environment for the efficient simulation of large-scale spiking neural networks on graphics processors[J].Neural Networks the Official Journal of the International Neural Network Society,2009,22(5-6):791-800.
[13]Brette R,Goodman D F.Vectorized algorithms for spiking neural network simulation[J].Neural Computation,2011,23(6):1503-1535.
[14]Dayan P,Abbott L F.Theoretical neuroscience[M].Massachusetts:MIT Press,2001:154-155.
[15]Brette R.Exact simulation of integrate-and-fire models with synaptic conductances[J].Neural Computation,2006,18(8):2004-2027.
[16]Hodgkin A L,Huxley A F.Currents carried by sodium and potassium ions through the membrane of the giant axon of Loligo[J].Journal of Physiology,1952,116(4):449.
[17]Li Lian-jiang.Research on GPU based spiking neural network learning[D].Wuhan:Huazhong University of Science and Technology,2015.(in Chinese)
[11]蔺想红.大规模脉冲神经网络的模拟与进化研究[D].哈尔滨:哈尔滨工业大学,2009.
[17]李连江.基于GPU的脉冲神经网络学习研究[D].武汉:华中科技大学,2015.