<div class="csl-bib-body">
<div class="csl-entry">Hanif, M. A., Sarda, G. M., Marchisio, A., Masera, G., Martina, M., & Shafique, M. (2022). CoNLoCNN: Exploiting Correlation and Non-Uniform Quantization for Energy-Efficient Low-precision Deep Convolutional Neural Networks. In <i>2Proceedings 2022 International Joint Conference on Neural Networks (IJCNN)</i> (pp. 1–8). https://doi.org/10.1109/IJCNN55064.2022.9892902</div>
</div>
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/150270
-
dc.description.abstract
In today's era of smart cyber-physical systems, Deep Neural Networks (DNNs) have become ubiquitous due to their state-of-the-art performance in complex real-world applications. The high computational complexity of these networks, which translates to increased energy consumption, is the foremost obstacle towards deploying large DNNs in resource-constrained systems. Fixed-Point (FP) implementations achieved through post-training quantization are commonly used to curtail the energy consumption of these networks. However, the uniform quantization intervals in FP restrict the bit-width of data structures to large values due to the need to represent most of the numbers with sufficient resolution and avoid high quantization errors. In this paper, we leverage the key insight that (in most of the scenarios) DNN weights and activations are mostly concentrated near zero and only a few of them have large magnitudes. We propose CoNLoCNN, a framework to enable energy-efficient low-precision deep convolutional neural network inference by exploiting: (1) non-uniform quantization of weights enabling simplification of complex multiplication operations; and (2) correlation between activation values enabling partial compensation of quantization errors at low cost without any run-time overheads. To significantly benefit from non-uniform quantization, we also propose a novel data representation format, Encoded Low-Precision Binary Signed Digit, to compress the bit-width of weights while ensuring direct use of the encoded weight for processing using a novel multiply-and-accumulate (MAC) unit design.
en
dc.language.iso
en
-
dc.subject
deep neural networks
en
dc.subject
quantization
en
dc.subject
energy efficiency
en
dc.subject
non-uniform quantization
en
dc.subject
Hardware Accelerator
en
dc.subject
low-precision hardware accelerator
en
dc.subject
error compensation
en
dc.title
CoNLoCNN: Exploiting Correlation and Non-Uniform Quantization for Energy-Efficient Low-precision Deep Convolutional Neural Networks
en
dc.type
Inproceedings
en
dc.type
Konferenzbeitrag
de
dc.contributor.affiliation
Politecnico di Torino
-
dc.contributor.affiliation
Politecnico di Torino
-
dc.contributor.affiliation
Politecnico di Torino
-
dc.relation.isbn
978-1-7281-8671-9
-
dc.description.startpage
1
-
dc.description.endpage
8
-
dc.type.category
Full-Paper Contribution
-
tuw.booktitle
2Proceedings 2022 International Joint Conference on Neural Networks (IJCNN)
-
tuw.container.volume
2022-July
-
tuw.peerreviewed
true
-
tuw.researchTopic.id
I2
-
tuw.researchTopic.name
Computer Engineering and Software-Intensive Systems
-
tuw.researchTopic.value
100
-
tuw.publication.orgunit
E191-02 - Forschungsbereich Embedded Computing Systems
-
tuw.publisher.doi
10.1109/IJCNN55064.2022.9892902
-
dc.description.numberOfPages
8
-
tuw.author.orcid
0000-0001-6231-3553
-
tuw.author.orcid
0000-0002-0689-4776
-
tuw.event.name
2022 International Joint Conference on Neural Networks (IJCNN)
en
tuw.event.startdate
18-07-2022
-
tuw.event.enddate
23-07-2022
-
tuw.event.online
Hybrid
-
tuw.event.type
Event for scientific audience
-
tuw.event.place
Padua
-
tuw.event.country
IT
-
tuw.event.presenter
Marchisio, Alberto
-
wb.sciencebranch
Informatik
-
wb.sciencebranch.oefos
1020
-
wb.sciencebranch.value
100
-
item.openairetype
Inproceedings
-
item.openairetype
Konferenzbeitrag
-
item.grantfulltext
restricted
-
item.cerifentitytype
Publications
-
item.cerifentitytype
Publications
-
item.languageiso639-1
en
-
item.openairecristype
http://purl.org/coar/resource_type/c_18cf
-
item.openairecristype
http://purl.org/coar/resource_type/c_18cf
-
item.fulltext
no Fulltext
-
crisitem.author.dept
E191-02 - Forschungsbereich Embedded Computing Systems
-
crisitem.author.dept
Politecnico di Torino
-
crisitem.author.dept
E191-02 - Forschungsbereich Embedded Computing Systems
-
crisitem.author.dept
Politecnico di Torino
-
crisitem.author.dept
Politecnico di Torino, Italy
-
crisitem.author.dept
E191-02 - Forschungsbereich Embedded Computing Systems