Neuromorphic extreme learning machines with bimodal memristive synapses

The biology-inspired intelligent computing system for the neuromorphic hardware implementation is useful in high-speed parallel information processing. However, traditional Von Neumann computer architecture and unsatisfactory signal transmission approach have jointly limited overall performance of specific implementation. In this paper, a compact extreme learning machine (ELM) synthesized with spintronic memristor-based synaptic circuit, biasing activation function circuit presented. Notably, due to threshold characteristic memristive device, has bimodal behavior. Namely, it capable provide constant adjustable network weights between adjacent layers ELM. Furthermore, two major limitations (process variations sneak path issue) are taken into account detailed robustness analysis whole network. Finally, entire scheme verified case studies single image super-resolution (SR) reconstruction.