A Survey of Bio‐Inspired and Other Alternative Architectures

摘要

Abstract Since the earliest days of the electronic computer, a small group of people has always seen the computer as an extension of biology, and endeavored to build computing models and even hardware which are inspired by, and in some cases direct copies of, biological systems. This chapter presents a brief survey of the specialized hardware developed over the past 20 years to support neurobiological models of computation. It includes a brief examination of current efforts and speculates on how such hardware, especially when implemented in nanoscale electronics, might offer unprecedented computing density, possibly leading to new capabilities in computational intelligence. The mix of continued Moore's law scaling, models from computational neuroscience, and molecular‐scale technology portends a potential paradigm shift in how computing is carried out. Among other things, the future of computing is most likely not about discrete logic, but rather about encoding, learning, and performing inference over stochastic variables. There may be a wide range of applications for such devices in robotics, and in the reduction and compression of widely distributed sensor data and power management.