Добірка наукової літератури з теми "Asynchronous Ballistic Reversible Computing"

Abstract The memristor-based neural crossbar is considered a promising device for research on neuromorphic computing. Moreover, memcapacitors can address the limitations caused by the resistive nature of memristors. A device with coexisting memristive and memcapacitive effects can provide rich features for neuromorphic computing systems. In this study, a device with a Pt/NiO x /NiO/Pt structure was developed; it demonstrates coexisting analog memristive and memcapacitive effects. The metallic NiO x serves as the oxygen storage layer and part of the top electrode. Analog memristive and memcapacitive effects are asynchronous; the resistance increases with reduction in capacitance. The physical mechanism for the asynchronous switching behavior of the resistance and capacitance was attributed to the reversible migration of oxygen ions between the metallic NiO x and semiconductor NiO in an electric field. The device was applied to mimicking simple synaptic functions such as short-term potentiation and short-term depression with small nonlinearity (0.0079 for potentiation and 0.00298 for depression). Complex neural properties such as homeostatic plasticity and habituation/sensitization were successfully simulated. This multifunctional device presents new applications for tunable oscillators and artificial neural networks (ANNs).