Emerging Computing Mechanisms for Edge AI

The aspiration of mimicking the human brain’s efficiency and smartness has brought notable progress in artificial intelligence (AI). While algorithms stemming from neural networks have successfully demonstrated their high performance across diverse applications, their ever-growing size and complexity urge to develop highly efficient hardware. Fortunately, the discovery of emerging memory devices, namely memristor, enables neuromorphic computing that mirrors computation methodologies in the brain. This paper reviews the motivations of neuromorphic computing and provides challenges and status quo in the memristor-based neuromorphic hardware designs, enfolding the progress in memristor devices, neuron and synaptic circuits, and neuromorphic systems. Specialized design for machine learning acceleration in neuromorphic systems, in-memory computing, is also discussed to envision a further higher leap of energy efficiency in computing hardware for the future AI.