Analog integrate-and-fire neurochips: neural competition in frequency and time domains

In this report, we present all inhibitory neural network, implemented on analog CMOS chips, that exhibits competitive behaviors in the frequency and time domains. The circuit for each neuron was designed to produce sequences in time of identically-shaped pulses, called spikes. The result of experiment and simulation revealed that the network more efficiently achieved the selective activation and inactivation of the neuron circuits on the basis of spike timing rather than of firing rates. The results indicate that the spike-timing-based neural processing by spiking neuron circuits provides a possible way of overcoming low tolerance problems of analog devices in noisy environments.