EdgeNAS: discovering efficient neural architectures for edge systems
Edge systems integrated with deep neural networks (DNNs) are deemed to pave the way for future artificial intelligence (AI). However, designing accurate and efficient DNNs for resource-limited edge systems is challenging as well as requires a huge amount of engineering efforts from human experts sin...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/165560 |
| _version_ | 1811676795753201664 |
|---|---|
| author | Luo, Xiangzhong Liu, Di Kong, Hao Liu, Weichen |
| author2 | School of Computer Science and Engineering |
| author_facet | School of Computer Science and Engineering Luo, Xiangzhong Liu, Di Kong, Hao Liu, Weichen |
| author_sort | Luo, Xiangzhong |
| collection | NTU |
| description | Edge systems integrated with deep neural networks (DNNs) are deemed to pave the way for future artificial intelligence (AI). However, designing accurate and efficient DNNs for resource-limited edge systems is challenging as well as requires a huge amount of engineering efforts from human experts since the design space is highly complex and diverse. Also, previous works mostly focus on designing DNNs with less floating-point operations (FLOPs), but indirect FLOPs count does not necessarily reflect the complexity of DNNs. To tackle these, we, in this paper, propose a novel neural architecture search (NAS) approach, namely EdgeNAS, to automatically discover efficient DNNs for less capable edge systems. To this end, we propose an end-to-end learning-based latency estimator, which is able to directly approximate the architecture latency on edge systems while incurring negligible computational overheads. Further, we effectively incorporate the latency estimator into EdgeNAS with a uniform sampling strategy, which guides the architecture search towards an edge-efficient direction. Moreover, a search space regularization approach is introduced to balance the trade-off between efficiency and accuracy. We evaluate EdgeNAS on the edge platform, Nvidia Jetson Xavier, with three popular datasets. Experimental results demonstrate the superiority of EdgeNAS over state-of-the-art approaches in terms of latency, accuracy, number of parameters, and the search cost. |
| first_indexed | 2024-10-01T02:27:09Z |
| format | Conference Paper |
| id | ntu-10356/165560 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T02:27:09Z |
| publishDate | 2023 |
| record_format | dspace |
| spelling | ntu-10356/1655602023-12-15T01:06:01Z EdgeNAS: discovering efficient neural architectures for edge systems Luo, Xiangzhong Liu, Di Kong, Hao Liu, Weichen School of Computer Science and Engineering 2020 IEEE 38th International Conference on Computer Design (ICCD) Parallel and Distributed Computing Centre Engineering::Computer science and engineering Neural Architecture Search Edge Systems Edge systems integrated with deep neural networks (DNNs) are deemed to pave the way for future artificial intelligence (AI). However, designing accurate and efficient DNNs for resource-limited edge systems is challenging as well as requires a huge amount of engineering efforts from human experts since the design space is highly complex and diverse. Also, previous works mostly focus on designing DNNs with less floating-point operations (FLOPs), but indirect FLOPs count does not necessarily reflect the complexity of DNNs. To tackle these, we, in this paper, propose a novel neural architecture search (NAS) approach, namely EdgeNAS, to automatically discover efficient DNNs for less capable edge systems. To this end, we propose an end-to-end learning-based latency estimator, which is able to directly approximate the architecture latency on edge systems while incurring negligible computational overheads. Further, we effectively incorporate the latency estimator into EdgeNAS with a uniform sampling strategy, which guides the architecture search towards an edge-efficient direction. Moreover, a search space regularization approach is introduced to balance the trade-off between efficiency and accuracy. We evaluate EdgeNAS on the edge platform, Nvidia Jetson Xavier, with three popular datasets. Experimental results demonstrate the superiority of EdgeNAS over state-of-the-art approaches in terms of latency, accuracy, number of parameters, and the search cost. Ministry of Education (MOE) Nanyang Technological University Submitted/Accepted version This work is partially supported by the Ministry of Education, Singapore, under its Academic Research Fund Tier 2 (MOE2019-T2-1-071) and Tier 1 (MOE2019-T1-001-072), and partially supported by Nanyang Technological University, Singapore, under its NAP (M4082282) and SUG (M4082087). 2023-03-31T05:11:57Z 2023-03-31T05:11:57Z 2020 Conference Paper Luo, X., Liu, D., Kong, H. & Liu, W. (2020). EdgeNAS: discovering efficient neural architectures for edge systems. 2020 IEEE 38th International Conference on Computer Design (ICCD), 288-295. https://dx.doi.org/10.1109/ICCD50377.2020.00056 https://hdl.handle.net/10356/165560 10.1109/ICCD50377.2020.00056 288 295 en MOE2019-T2-1-071 MOE2019-T1- 001-072 NAP (M4082282) SUG (M4082087) 10.21979/N9/L2QVIV © 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The published version is available at: https://doi.org/10.1109/ICCD50377.2020.00056. application/pdf |
| spellingShingle | Engineering::Computer science and engineering Neural Architecture Search Edge Systems Luo, Xiangzhong Liu, Di Kong, Hao Liu, Weichen EdgeNAS: discovering efficient neural architectures for edge systems |
| title | EdgeNAS: discovering efficient neural architectures for edge systems |
| title_full | EdgeNAS: discovering efficient neural architectures for edge systems |
| title_fullStr | EdgeNAS: discovering efficient neural architectures for edge systems |
| title_full_unstemmed | EdgeNAS: discovering efficient neural architectures for edge systems |
| title_short | EdgeNAS: discovering efficient neural architectures for edge systems |
| title_sort | edgenas discovering efficient neural architectures for edge systems |
| topic | Engineering::Computer science and engineering Neural Architecture Search Edge Systems |
| url | https://hdl.handle.net/10356/165560 |
| work_keys_str_mv | AT luoxiangzhong edgenasdiscoveringefficientneuralarchitecturesforedgesystems AT liudi edgenasdiscoveringefficientneuralarchitecturesforedgesystems AT konghao edgenasdiscoveringefficientneuralarchitecturesforedgesystems AT liuweichen edgenasdiscoveringefficientneuralarchitecturesforedgesystems |