improved unsupervised clustering by spare temporal coding and network of RBF neurons

temporal importance of pulses generated by neurons is evidenced more than before and there are efforts to use these pulses for information encoding. Extending on hopfield's idea. Natschlager&ruf proposed a learning algorithm that performs unsupervised clustering in spiling neural networks using spike times as input . this model is limited in both cluster capacity and precision then bohte poutre and kok overcame this problem using receptive fields idea. in this paper our goal is to improve the bohte's model so that the precision of the network will be increased. a critical issue that has not been addressed in previous models is the number of synopsis terminals between the input layer neurons and each of the output layer neurons.

کلیدواژه ها:

spiking neurons ، clustering ، RBFneurons