Phonetic Classification with Spiking Neural Network Using a Gradient Descent Rule

Being the closest model of the biological neuron, the spiking neuron is the third and newest generation of artificial neuron. The particularity of this neuron is the use of temporal coding to pass information between network units. Using such codes allows the transmission of a large amount of data with only few spikes, simply one or zero for each neuron involved in the specific processing task. The true deal is how to encode analogical information to a spikes train. More, it’s not the only problem which we find in using spiking neurons network (SNN), we have to choose different parameters and functions. In this paper, in the middle of several spiking neurons models, we have chosen the spiking response model (SRM) to apply in phonetic classification using phonemes from TIMIT databases. Before, for the studies, we have performed experiments for the classical Xor-problem and explore the impact of encoding information on the network structure. The learning rules used in this experiment was based on error backpropagation based on time to first spike.