A mixed-signal universal neuromorphic computing system

摘要

Neuromorphic information processing systems offer the potential to overcome imminent problems of state-of-the-art computers, in particular the energy efficiency problem, the device reliability problem and the software complexity problem. This paper starts with a short overview of state-of-the-art neuromorphic hardware implementations and their applications. It then describes the time-accelerated mixed-signal approach of the BrainScaleS project in some detail.