DoubleQExt: Hardware and Memory Efficient CNN Through Two Levels of Quantization
DC Field | Value | Language |
---|---|---|
dc.contributor.author | See, Jin-Chuan | - |
dc.contributor.author | Ng, Hui-Fuang | - |
dc.contributor.author | Tan, Hung-Khoon | - |
dc.contributor.author | Chang, Jing-Jing | - |
dc.contributor.author | Lee, Wai-Kong | - |
dc.contributor.author | Hwang, Seong Oun | - |
dc.date.accessioned | 2022-01-08T06:40:50Z | - |
dc.date.available | 2022-01-08T06:40:50Z | - |
dc.date.created | 2022-01-08 | - |
dc.date.issued | 2021-12 | - |
dc.identifier.issn | 2169-3536 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/gachon/handle/2020.sw.gachon/83190 | - |
dc.description.abstract | To fulfil the tight area and memory constraints in IoT applications, the design of efficient Convolutional Neural Network (CNN) hardware becomes crucial. Quantization of CNN is one of the promising approach that allows the compression of large CNN into a much smaller one, which is very suitable for IoT applications. Among various proposed quantization schemes, Power-of-two (PoT) quantization enables efficient hardware implementation and small memory consumption for CNN accelerators, but requires retraining of CNN to retain its accuracy. This paper proposes a two-level post-training static quantization technique (DoubleQ) that combines the 8-bit and PoT weight quantization. The CNN weight is first quantized to 8-bit (level one), then further quantized to PoT (level two). This allows multiplication to be carried out using shifters, by expressing the weights in their PoT exponent form. DoubleQ also reduces the memory storage requirement for CNN, as only the exponent of the weights is needed for storage. However, DoubleQ trades the accuracy of the network for reduced memory storage. To recover the accuracy, a selection process (DoubleQExt) was proposed to strategically select some of the less informative layers in the network to be quantized with PoT at the second level. On ResNet-20, the proposed DoubleQ can reduce the memory consumption by 37.50% with 7.28% accuracy degradation compared to 8-bit quantization. By applying DoubleQExt, the accuracy is only degraded by 1.19% compared to 8-bit version while achieving a memory reduction of 23.05%. This result is also 1% more accurate than the state-of-the-art work (SegLog). The proposed DoubleQExt also allows flexible configuration to trade off the memory consumption with better accuracy, which is not found in the other state-of-the-art works. With the proposed two-level weight quantization, one can achieve a more efficient hardware architecture for CNN with minimal impact to the accuracy, which is crucial for IoT applications. | - |
dc.language | 영어 | - |
dc.language.iso | en | - |
dc.publisher | IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC | - |
dc.relation.isPartOf | IEEE ACCESS | - |
dc.title | DoubleQExt: Hardware and Memory Efficient CNN Through Two Levels of Quantization | - |
dc.type | Article | - |
dc.type.rims | ART | - |
dc.description.journalClass | 1 | - |
dc.identifier.wosid | 000736733200001 | - |
dc.identifier.doi | 10.1109/ACCESS.2021.3138756 | - |
dc.identifier.bibliographicCitation | IEEE ACCESS, v.9, pp.169082 - 169091 | - |
dc.description.isOpenAccess | N | - |
dc.identifier.scopusid | 2-s2.0-85122326728 | - |
dc.citation.endPage | 169091 | - |
dc.citation.startPage | 169082 | - |
dc.citation.title | IEEE ACCESS | - |
dc.citation.volume | 9 | - |
dc.contributor.affiliatedAuthor | Lee, Wai-Kong | - |
dc.contributor.affiliatedAuthor | Hwang, Seong Oun | - |
dc.type.docType | Article | - |
dc.subject.keywordAuthor | Quantization (signal) | - |
dc.subject.keywordAuthor | Convolutional neural networks | - |
dc.subject.keywordAuthor | Hardware | - |
dc.subject.keywordAuthor | Memory management | - |
dc.subject.keywordAuthor | Field programmable gate arrays | - |
dc.subject.keywordAuthor | Internet of Things | - |
dc.subject.keywordAuthor | Degradation | - |
dc.subject.keywordAuthor | Convolutional neural network | - |
dc.subject.keywordAuthor | quantization | - |
dc.subject.keywordAuthor | Internet of Things | - |
dc.subject.keywordAuthor | deep learning | - |
dc.subject.keywordAuthor | field programmable gate array | - |
dc.subject.keywordPlus | CONVOLUTIONAL NEURAL-NETWORKS | - |
dc.subject.keywordPlus | REAL-TIME | - |
dc.relation.journalResearchArea | Computer Science | - |
dc.relation.journalResearchArea | Engineering | - |
dc.relation.journalResearchArea | Telecommunications | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Information Systems | - |
dc.relation.journalWebOfScienceCategory | Engineering, Electrical & Electronic | - |
dc.relation.journalWebOfScienceCategory | Telecommunications | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
1342, Seongnam-daero, Sujeong-gu, Seongnam-si, Gyeonggi-do, Republic of Korea(13120)031-750-5114
COPYRIGHT 2020 Gachon University All Rights Reserved.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.