DC Field | Value | Language |
---|---|---|
dc.contributor.author | Kim, Han-Byul | - |
dc.contributor.author | Park, Eunhyeok | - |
dc.contributor.author | Yoo, Sungjoo | - |
dc.date.accessioned | 2023-03-06T00:22:42Z | - |
dc.date.available | 2023-03-06T00:22:42Z | - |
dc.date.created | 2023-03-03 | - |
dc.date.issued | 2022-10-25 | - |
dc.identifier.uri | https://oasis.postech.ac.kr/handle/2014.oak/116834 | - |
dc.description.abstract | In this paper, we propose Branch-wise Activation-clipping Search Quantization (BASQ), which is a novel quantization method for low-bit activation. BASQ optimizes clip value in continuous search space while simultaneously searching L2 decay weight factor for updating clip value in discrete search space. We also propose a novel block structure for low precision that works properly on both MobileNet and ResNet structures with branch-wise searching. We evaluate the proposed methods by quantizing both weights and activations to 4-bit or lower. Contrary to the existing methods which are effective only for redundant networks, e.g., ResNet-18, or highly optimized networks, e.g., MobileNet-v2, our proposed method offers constant competitiveness on both types of networks across low precisions from 2 to 4-bits. Specifically, our 2-bit MobileNet-v2 offers top-1 accuracy of 64.71% on ImageNet, outperforming the existing method by a large margin (2.8%), and our 4-bit MobileNet-v2 gives 71.98% which is comparable to the full-precision accuracy 71.88% while our uniform quantization method offers comparable accuracy of 2-bit ResNet-18 to the state-of-the-art non-uniform quantization method. Source code is on https://github.com/HanByulKim/BASQ. | - |
dc.language | English | - |
dc.publisher | Springer Science and Business Media Deutschland GmbH | - |
dc.relation.isPartOf | 17th European Conference on Computer Vision, ECCV 2022 | - |
dc.relation.isPartOf | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | - |
dc.title | BASQ: Branch-wise Activation-clipping Search Quantization for Sub-4-bit Neural Networks | - |
dc.type | Conference | - |
dc.type.rims | CONF | - |
dc.identifier.bibliographicCitation | 17th European Conference on Computer Vision, ECCV 2022, pp.17 - 33 | - |
dc.identifier.wosid | 000897093900002 | - |
dc.citation.conferenceDate | 2022-10-23 | - |
dc.citation.conferencePlace | IS | - |
dc.citation.endPage | 33 | - |
dc.citation.startPage | 17 | - |
dc.citation.title | 17th European Conference on Computer Vision, ECCV 2022 | - |
dc.contributor.affiliatedAuthor | Park, Eunhyeok | - |
dc.identifier.scopusid | 2-s2.0-85142743806 | - |
dc.description.journalClass | 1 | - |
dc.description.journalClass | 1 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
library@postech.ac.kr Tel: 054-279-2548
Copyrights © by 2017 Pohang University of Science ad Technology All right reserved.