CHIH-I WU2021-09-022021-09-02202007431562https://www.scopus.com/inward/record.uri?eid=2-s2.0-85098172031&doi=10.1109%2fVLSITechnology18217.2020.9265033&partnerID=40&md5=3dfd1022e06122f052d241ee98567f87https://scholars.lib.ntu.edu.tw/handle/123456789/580696A unique compact Poisson neuron that encodes information in the tunable duty cycle of probabilistic spike trains is presented as an enabling technology for cost-effective spiking neural network (SNN) hardware. The Poisson neuron exploits the back-hopping oscillation (BHO) in scalable spin-transfer torque (STT)-MRAM. The macrospin LLGS simulation confirms that the coupled local Joule heating and STT effects are responsible for the bias-dependent BHO. The complete neuron circuit design is at least 6*smaller than the state-of-the-art integrate-and- fire (IF) CMOS neuron. Hardware-friendly all-spin deep SNNs achieve equivalent accuracy to deep neural networks (DNN), 98.4 % for MNIST, even when considering the probabilistic nature of neurons. ? 2020 IEEE.Cost effectiveness; Deep neural networks; Magnetic recording; Neurons; VLSI circuits; Cost effective; Enabling technologies; Integrate and fires; Neuron circuits; Spiking neural network(SNN); Spiking neural networks; Spin transfer torque; State of the art; Neural networks[SDGs]SDG7conference paper10.1109/VLSITechnology18217.2020.92650332-s2.0-85098172031