Memristive dynamics enabled neuromorphic computing systems

Abstract

The slowing down of transistor scaling and explosive growth for intelligence computing power emerge as the two driving factors for the study of novel devices and materials to pursue highly-efficient computing systems. Memristors, incorporating rich intrinsic dynamics, are a promising candidate for constructing efficient and scalable bio-inspired computing systems. In this progress report, we review the latest advances in novel types of memristors as well as their applications in implementing neuromorphic computing systems. This paper not only covers the memristive dynamics-enabled bionic computing systems but also discusses the memristive sensory systems that integrate sensing and computing. Eventually, device-circuit co-optimization methods are given to emphasize the trend of cross-layer co-design in this fast-evolving field. The innovation in memristor devices mainly focuses on specialized computing hardware and yields superior computing and sensing efficiency. At last, we offer our insight into the trend of state-of-the-art research in memristive materials, devices, circuits, and systems.