Charge-based Neuromorphic Cell by InGaZnO Transistor and Implementation of Simple Scheme Spike-Timing-Dependent Plasticity

This paper presents a novel charge-based neuromorphic cell architecture composed of 1-capacitor and 1-transistor using extremely low off leakage (<10(-21)A/mu m) oxide-semiconductor transistors. Thanks to ultra-long retention characteristics, circuit implementation of spike-timing-dependent plasticity, for spiking neural network, were successfully demonstrated. Cell array architecture and it operation were confirmed by simulation with charging and discharging control. Asynchronous STDP implementation was also achieved without complicated forming of input pulse, and low-power operation less than similar to 0.1pJ can be expected.