神经形态工程学
仿真
冯·诺依曼建筑
能源消耗
突触
计算机科学
能量(信号处理)
人工神经网络
人工智能
计算机体系结构
神经科学
材料科学
消费(社会学)
工程类
电气工程
物理
生物
操作系统
量子力学
经济增长
经济
作者
Yeongjun Lee,Hea-Lim Park,Yeong-In Kim,Takhee Lee
出处
期刊:Joule
[Elsevier]
日期:2021-04-01
卷期号:5 (4): 794-810
被引量:37
标识
DOI:10.1016/j.joule.2021.01.005
摘要
Summary
The von Neumann computing architecture consists of separated processing and memory elements; it is too bulky and energy-intensive to be implemented in the upcoming artificial intelligence age. In contrast, neurons and synapses in a brain perform learning and memory in an integrated manner and function energy-efficiently by analog adjustment of synaptic strengths in response to stimulation. Organic artificial synapses provide good emulation of the functions and structures of biological synapses and are easily fabricated and therefore can be applied to various neuromorphic electronic devices. In particular, organic artificial synapses that consume energy at a level comparable to that of a biological synapse show great promise for use in future low-energy neuromorphic devices. Here, we review the trends of energy consumption of organic artificial synapses and how it is affected by the structure, materials, and operation mechanism. We also present a strategy to decrease the energy consumption of organic neuromorphic devices. Our review will help the development of versatile low-energy organic neuromorphic electronics.
科研通智能强力驱动
Strongly Powered by AbleSci AI