Materials challenges and opportunities for brain-inspired computing

Inspired by the working principles of the human brain, neuromorphic computing shows great potential in executing cognitive tasks such as learning and adaptation with high energy efficiency. A major challenge is the development of devices and circuits that can naturally replicate the behavior of neur...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:MRS bulletin Ročník 46; číslo 10; s. 978 - 986
Hlavní autoři: Zhao, Y. D., Kang, J. F., Ielmini, D.
Médium: Journal Article
Jazyk:angličtina
Vydáno: Cham Springer International Publishing 01.10.2021
Springer Nature B.V
Témata:
Applied and Technical Physics
Brain
Characterization and Evaluation of Materials
Chemistry and Materials Science
Circuits
Cognitive tasks
Energy consumption
Energy Materials
Hardware
Materials Engineering
Materials Science
Nanotechnology
Neuromorphic computing
Review Article
Synapses
ISSN:0883-7694, 1938-1425
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Inspired by the working principles of the human brain, neuromorphic computing shows great potential in executing cognitive tasks such as learning and adaptation with high energy efficiency. A major challenge is the development of devices and circuits that can naturally replicate the behavior of neurons and synapses, thus reducing the complexity, the energy consumption, and the area of the neuromorphic chip. Recently, much progress has been achieved in realizing hardware neuromorphic circuits with emerging “memristive” materials and devices, which present a wealth of physical phenomena that appear promising for the ad hoc design of virtually any neuromorphic function in the scale of few square nanometers on a silicon chip. In this article, an overview of material opportunities on emerging devices for brain-inspired computing is provided. We will summarize the biological functions of neuromorphic elements, discuss the requirements for their material counterparts and review the recent progress, and illustrate some cognitive computing primitives in hardware networks. Finally, the upcoming materials challenges will be discussed. Graphic abstract
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0883-7694
1938-1425
DOI:10.1557/s43577-021-00205-1