Committee Machines—A Universal Method to Deal with Non-Idealities in Memristor-Based Neural Networks
- Submitting institution
-
Liverpool John Moores University
- Unit of assessment
- 12 - Engineering
- Output identifier
- 1978
- Type
- D - Journal article
- DOI
-
10.1038/s41467-020-18098-0
- Title of journal
- Nature Communications
- Article number
- 4273
- First page
- -
- Volume
- 11
- Issue
- 1
- ISSN
- 2041-1723
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2020
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
9
- Research group(s)
-
D - RCEEE
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Key output from EPSRC grants (EP/S000259/1, EP/S000224/1, £786k, Lead PI: Wei Zhang, 2018-2021; EP/P013503/1, £735k, 2017-2020) and the Leverhulme Trust (RPG-2016-135, 2016-2018, £331k). A new technique developed to overcome one of the most challenging issues in memristor-based neural networks by suppressing the impact of faulty devices, variations, and noises, for improved pattern recognition accuracy without even increasing the number of memristors. Produced in collaboration with UCL, UMass (USA), and world-leading research consortium IMEC (2015–2018, €300k) who is partnered with and disseminated the results to industrial companies including Intel, Micron, and Samsung (Dr. Gouri Kar, Memory Director, Gouri.Kar@imec.be).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -