The double queue method: a numerical method for integrate-and-fire neuron networks
| Autor: | Geehyuk Lee, Nabil H. Farhat |
|---|---|
| Rok vydání: | 2001 |
| Předmět: |
Neurons
Mathematical optimization Artificial neural network Differential equation Computer science Cognitive Neuroscience Numerical analysis Models Neurological Finite difference method Action Potentials Reproducibility of Results Classification of discontinuities Synaptic Transmission Synchronization Self-organized criticality Artificial Intelligence Animals Humans Neural Networks Computer Cortical Synchronization Nerve Net Queue Algorithm Algorithms |
| Zdroj: | Neural Networks. 14:921-932 |
| ISSN: | 0893-6080 |
| DOI: | 10.1016/s0893-6080(01)00034-x |
| Popis: | Numerical methods for initial-value problems based on finite-differencing of differential equations (FDM) are not well suited for the simulation of an integrate-and-fire neuron network (IFNN) due to the discontinuities implied by the firing condition of the neurons. The Double Queue Method (DQM) is an event-queue based numerical method designed for the simulation of an IFNN that can deal with such discontinuities properly. In the DQM, the states of individual neurons at the next predicted discontinuous points are determined by an analytic solution, meaning an optimal performance in both accuracy and speed. A comparison study with the FDM demonstrates the superiority of the DQM, and provides some examples where the FDM gives inaccurate results that can possibly lead to a false conclusion about the dynamics of an IFNN. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect