HIGHLIGHTS
SUMMARY
Spiking Neural_Networks (SNNs) are an alternative to sigmoidal neural_networks, which add the concept of time into their operations. This third class of artificial neural_networks consists of neurons that transmit information employing discrete impulses known as spikes. Supervised gradient-based training algorithms for sigmoidal neural_networks can not directly be employed for SNNs due to the discrete nature of spikes. Several methods have used STDP to train spiking neural_networks for classification problems. After training each layer, they freeze its synaptic weights and disable its inhibitory strategies/neurons to provide enough spikes (information) to train the next . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.