Português
Ciência-IUL
Publications
Publication Detailed Description
ESANN 2022 proceedings
Year (definitive publication)
2022
Language
English
Country
--
More Information
Web of Science®
This publication is not indexed in Web of Science®
Scopus
This publication is not indexed in Scopus
Google Scholar
Abstract
Weightless neural networks (WNNs) are a type of machine learning model which perform prediction using lookup tables (LUTs) instead of arithmetic operations. Recent advancements in WNNs have reduced model sizes and improved accuracies, reducing the gap in accuracy with deep neural networks (DNNs). Modern DNNs leverage “pruning” techniques to reduce model size, but this has not previously been explored for WNNs. We propose a WNN pruning strategy based on identifying and culling the LUTs which contribute least to overall model accuracy. We demonstrate an average 40% reduction in model size with at most 1% reduction in accuracy.
Acknowledgements
--
Keywords
Fields of Science and Technology Classification
- Computer and Information Sciences - Natural Sciences
Funding Records
| Funding Reference | Funding Entity |
|---|---|
| 3015.001/3016.00 | Semiconductor Research Corporation |
| POCI-01-0247-FEDER-045912 | FEDER |
| UIDP/04466/2020 | Fundação para a Ciência e a Tecnologia |
| UIDB/04466/2020 | Fundação para a Ciência e a Tecnologia |
Related Projects
This publication is an output of the following project(s):
