Wave-Based Neuromorphic Computing Framework for Brain-Like Energy Efficiency and Integration

We present a framework of wave-based neuromorphic computing aiming at brain-like capabilities and efficiencies with nanoscale device integration. We take advantage of the unique nature of elastic nondissipative wave dynamics in both computations and IO communications in between as a means to natively implement and execute neuromorphic computing functions such as weighted sum in a spatiotemporal manner. Lower bound analysis based on a memory model and wave group velocity scaling is provided for conceptual evaluations.