Publications
Detailed Information
A novel zero weight/activation-aware hardware architecture of convolutional neural network
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Kim, D. | - |
dc.contributor.author | Ahn, J. | - |
dc.contributor.author | Yoo, S. | - |
dc.date.accessioned | 2023-06-27T06:37:42Z | - |
dc.date.available | 2023-06-27T06:37:42Z | - |
dc.date.created | 2023-06-19 | - |
dc.date.issued | 2017-05 | - |
dc.identifier.citation | Proceedings of the 2017 Design, Automation and Test in Europe, DATE 2017, pp.1462-1467 | - |
dc.identifier.issn | 0000-0000 | - |
dc.identifier.uri | https://hdl.handle.net/10371/192892 | - |
dc.description.abstract | It is imperative to accelerate convolutional neural networks (CNNs) due to their ever-widening application areas from server, mobile to IoT devices. Based on the fact that CNNs can be characterized by a significant amount of zero values in both kernel weights and activations, we propose a novel hardware accelerator for CNNs exploiting zero weights and activations. We also report a zero-induced load imbalance problem, which exists in zero-aware parallel CNN hardware architectures, and present a zero-aware kernel allocation as a solution. According to our experiments with a cycle-accurate simulation model, RTL, and layout design of the proposed architecture running two real deep CNNs, pruned AlexNet [1] and VGG-16 [2], our architecture offers 4x/1.8x (AlexNet) and 5.2x/2.1x (VGG-16) speedup compared with state-of-the-art zero-agnostic/zero-activation-aware architectures. © 2017 IEEE. | - |
dc.language | 영어 | - |
dc.publisher | Institute of Electrical and Electronics Engineers Inc. | - |
dc.title | A novel zero weight/activation-aware hardware architecture of convolutional neural network | - |
dc.type | Article | - |
dc.citation.journaltitle | Proceedings of the 2017 Design, Automation and Test in Europe, DATE 2017 | - |
dc.identifier.scopusid | 2-s2.0-85020166363 | - |
dc.citation.endpage | 1467 | - |
dc.citation.startpage | 1462 | - |
dc.description.isOpenAccess | N | - |
dc.contributor.affiliatedAuthor | Yoo, S. | - |
dc.description.journalClass | 1 | - |
- Appears in Collections:
- Files in This Item:
- There are no files associated with this item.
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.