DC Field | Value | Language |
---|---|---|
dc.contributor.author | Shin, Jaekang | ko |
dc.contributor.author | Choi, Seungkyu | ko |
dc.contributor.author | Ra, Jongwoo | ko |
dc.contributor.author | Kim, Lee-Sup | ko |
dc.date.accessioned | 2022-09-29T01:00:16Z | - |
dc.date.available | 2022-09-29T01:00:16Z | - |
dc.date.created | 2022-09-27 | - |
dc.date.created | 2022-09-27 | - |
dc.date.created | 2022-09-27 | - |
dc.date.issued | 2022-07 | - |
dc.identifier.citation | 59th ACM/IEEE Design Automation Conference, DAC 2022, pp.253 - 258 | - |
dc.identifier.issn | 0738-100X | - |
dc.identifier.uri | http://hdl.handle.net/10203/298761 | - |
dc.description.abstract | Real-world AI applications, such as augmented reality or autonomous driving, require processing multiple CV tasks simultaneously. However, the enormous data size and the memory footprint have been a crucial hurdle for deep neural networks to be applied in resource-constrained devices. To solve the problem, we propose an algorithm/architecture co-design. The proposed algorithmic scheme, named SqueeD, reduces per-task weight and activation size by 21.9x and 2.1x, respectively, by sharing those data between tasks. Moreover, we design architecture and dataflow to minimize DRAM access by fully utilizing benefits from SqueeD. As a result, the proposed architecture reduces the DRAM access increment and energy consumption increment per task by 2.2x and 1.3x, respectively. | - |
dc.language | English | - |
dc.publisher | Institute of Electrical and Electronics Engineers Inc. | - |
dc.title | Algorithm/architecture co-design for energy-efficient acceleration of multi-task DNN | - |
dc.type | Conference | - |
dc.identifier.scopusid | 2-s2.0-85137435799 | - |
dc.type.rims | CONF | - |
dc.citation.beginningpage | 253 | - |
dc.citation.endingpage | 258 | - |
dc.citation.publicationname | 59th ACM/IEEE Design Automation Conference, DAC 2022 | - |
dc.identifier.conferencecountry | US | - |
dc.identifier.conferencelocation | San Francisco | - |
dc.identifier.doi | 10.1145/3489517.3530455 | - |
dc.contributor.localauthor | Kim, Lee-Sup | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.