Power-efficient neural network with artificial dendrites

• Published in: Nature nanotechnology Vol. 15; no. 9; pp. 776 - 782
• Main Authors: Li, Xinyi, Tang, Jianshi, Zhang, Qingtian, Gao, Bin, Yang, J. Joshua, Song, Sen, Wu, Wei, Zhang, Wenqiang, Yao, Peng, Deng, Ning, Deng, Lei, Xie, Yuan, Qian, He, Wu, Huaqiang
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 01.09.2020; Nature Publishing Group
• Subjects: 639/166/987; 639/925/927/1007; Animals; Application specific integrated circuits; Artificial Cells; Artificial neural networks; Background noise; Central processing units; Chemistry and Materials Science; Computer simulation; CPUs; Databases, Factual; Dendrites; Dendrites - physiology; Electronics; Energy efficiency; Equipment Design; Image Processing, Computer-Assisted; Integrated circuits; Materials Science; Memristors; Mice; Models, Neurological; Multilayers; Nanotechnology; Nanotechnology and Microengineering; Nervous system; Neural networks; Neural Networks, Computer; Neurons - physiology; Oxygen - chemistry; Power consumption; Power management; Signal processing; Synapses; Task complexity
• ISSN: 1748-3387; 1748-3395; 1748-3395
• DOI: 10.1038/s41565-020-0722-5

In the nervous system, dendrites, branches of neurons that transmit signals between synapses and soma, play a critical role in processing functions, such as nonlinear integration of postsynaptic signals. The lack of these critical functions in artificial neural networks compromises their performance, for example in terms of flexibility, energy efficiency and the ability to handle complex tasks. Here, by developing artificial dendrites, we experimentally demonstrate a complete neural network fully integrated with synapses, dendrites and soma, implemented using scalable memristor devices. We perform a digit recognition task and simulate a multilayer network using experimentally derived device characteristics. The power consumption is more than three orders of magnitude lower than that of a central processing unit and 70 times lower than that of a typical application-specific integrated circuit chip. This network, equipped with functional dendrites, shows the potential of substantial overall performance improvement, for example by extracting critical information from a noisy background with significantly reduced power consumption and enhanced accuracy. A memristor-based artificial dendrite enables the neural network to perform high-accuracy computation tasks with reduced power consumption.