Domain Wall Leaky Integrate-and-Fire Neurons With Shape-Based Configurable Activation Functions

Complementary metal oxide semiconductor (CMOS) devices display volatile characteristics, and are not well suited for analog applications such as neuromorphic computing. Spintronic devices, on the other hand, exhibit both non-volatile features, which well-suited to Consequently, these novel at forefront of beyond-CMOS artificial intelligence applications. However, a large quantity still require use CMOS, decreases efficiency system. To resolve this, we have previously proposed number neurons synapses that do CMOS operation. Although significant improvement over previous renditions, their ability enable neural network learning recognition is limited by intrinsic activation functions. This work proposes modifications spintronic configuration functions through control shape magnetic domain wall track. Linear sigmoidal demonstrated in this work, can be extended similar approach wide variety