Paper Info
Title
Mixing circuit based on neural associative memories and nanoelectronic 1S1R cells
Abstract
A binary neural associative memory concept comprising nanoelectronic resistive switches and non-linear selector devices used as binary connection elements is proposed for the realization of logic and arithmetic functions. Based on the consideration of device variability an error detecting 2-out-of-n-code for the representation of operands is developed. The connection structure of an associative memory is derived which implements a problem-independent mixing operation which is seen as an elementary operation for mapping operands to desired symbols. The mixing operation maps two independent 2-out-of-n-encoded operands to a unique 2-out-of-N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A</sub> -encoded address in an error-free way. Finally, it is shown that for selected subsets of the proposed 2-out-of-n code set a significant compaction of the connection matrix is possible.
Year
DOI
Venue
2017
10.1109/NANOARCH.2017.8053707
2017 IEEE/ACM International Symposium on Nanoscale Architectures (NANOARCH)
Keywords
Field
DocType
resistive switches,selector device,binary neural associative memory,BiNAM,clipped hebbian synaptic rule,variability,reliability,logic functions,arithmetic function,nanoelectronics,m-out-of-n code,constant-weight-code
Arithmetic function,Associative property,Content-addressable memory,Matrix (mathematics),Computer science,Resistive touchscreen,Operand,Parallel computing,Electronic engineering,Binary number
Conference
ISSN
ISBN
Citations 
2327-8218
978-1-5090-6038-2
0
PageRank 
References 
Authors
0.34
4
2
Name
Order
Citations
PageRank
Arne Heittmann1245.94
Tobias G. Noll219937.51