Premium
Memristive Devices and Networks for Brain‐Inspired Computing
Author(s) -
Zhang Teng,
Yang Ke,
Xu Xiaoyan,
Cai Yimao,
Yang Yuchao,
Huang Ru
Publication year - 2019
Publication title -
physica status solidi (rrl) – rapid research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.786
H-Index - 68
eISSN - 1862-6270
pISSN - 1862-6254
DOI - 10.1002/pssr.201970031
Subject(s) - neuromorphic engineering , computer science , memristor , artificial neural network , exploit , field (mathematics) , mechanism (biology) , perspective (graphical) , computer architecture , artificial intelligence , electronic engineering , engineering , physics , computer security , mathematics , quantum mechanics , pure mathematics
In the Review@RRL by Yuchao Yang, Ru Huang and co‐workers (article no. 1900029 ), existing approaches for the implementation of artificial synapses based on metal ion transport, oxygen ion transport and phase change mechanism, etc. are overviewed, and their advantages and disadvantages are analyzed. This is followed by latest progresses on memristive neurons, ranging from leaky‐integrate‐and‐fire neuron to Hodgkin‐Huxley neuron. From a system perspective, both neural network accelerators that exploit the in‐memory and analog characteristics of memristive arrays as well as neuromorphic systems based on the intrinsic dynamics of memristors are discussed. The authors finally point out the outstanding challenges this field faces and highlight the importance of co‐optimization between devices, circuits and algorithms.