Neuromorphic Dendritic Synapse Integrating Gated-RRAM

Artificial neural networks (ANNs) are becoming increasingly widespread yet the current hardware implementations exceed the power, area, and time budget of many applications. Additionally, the current ANN neurons are a highly simplified model of their biological counterparts. In this work, we propose a neuromorphic dendritic synapse that improves the computational complexity of individual neurons as a foundation for more efficient implementation of complex ANNs.