Weights Convergence and Spikes Correlation in an Adaptive Neural Network Implemented on VLSI
Résumé
This paper presents simulations of a conductance-based neural network implemented on a mixed hardwaresoftware simulation system. Synaptic connections follow a bio-realistic STDP rule. Neurons receive correlated input noise patterns, resulting in a weights convergence in a confined range of conductance values. The correlation of the output spike trains depends on the correlation degree of the input patterns