{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T06:38:53Z","timestamp":1790836733439,"version":"4.1.0"},"reference-count":47,"publisher":"Association for Computing Machinery (ACM)","issue":"5s","license":[{"start":{"date-parts":[[2023,9,9]],"date-time":"2023-09-09T00:00:00Z","timestamp":1694217600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Embed. Comput. Syst."],"published-print":{"date-parts":[[2023,10,31]]},"abstract":"<jats:p>\n                    Graph Neural Networks (GNNs) are becoming increasingly popular for vision-based applications due to their intrinsic capacity in modeling structural and contextual relations between various parts of an image frame. On another front, the rising popularity of deep vision-based applications at the edge has been facilitated by the recent advancements in heterogeneous multi-processor Systems on Chips (MPSoCs) that enable inference under real-time, stringent execution requirements. By extension, GNNs employed for vision-based applications must adhere to the same execution requirements. Yet contrary to typical deep neural networks, the irregular flow of graph learning operations poses a challenge to running GNNs on such heterogeneous MPSoC platforms. In this paper, we propose a novel unified\n                    <jats:italic>design-mapping<\/jats:italic>\n                    approach for efficient processing of vision GNN workloads on heterogeneous MPSoC platforms. Particularly, we develop MaGNAS, a mapping-aware Graph Neural Architecture Search framework. MaGNAS proposes a GNN architectural design space coupled with prospective mapping options on a heterogeneous SoC to identify model architectures that maximize on-device resource efficiency. To achieve this, MaGNAS employs a two-tier evolutionary search to identify optimal\n                    <jats:italic>GNNs<\/jats:italic>\n                    and\n                    <jats:italic>mapping<\/jats:italic>\n                    pairings that yield the best performance trade-offs. Through designing a supernet derived from the recent Vision GNN (ViG) architecture, we conducted experiments on four (04) state-of-the-art vision datasets using both (\n                    <jats:italic>i<\/jats:italic>\n                    ) a real hardware SoC platform (NVIDIA Xavier AGX) and (\n                    <jats:italic>ii<\/jats:italic>\n                    ) a performance\/cost model simulator for DNN accelerators. Our experimental results demonstrate that MaGNAS is able to provide\n                    <jats:bold>1.57<\/jats:bold>\n                    \u00d7 latency speedup and is\n                    <jats:bold>3.38<\/jats:bold>\n                    \u00d7 more energy-efficient for several vision datasets executed on the Xavier MPSoC vs. the GPU-only deployment while sustaining an average\n                    <jats:bold>0.11%<\/jats:bold>\n                    accuracy reduction from the baseline.\n                  <\/jats:p>","DOI":"10.1145\/3609386","type":"journal-article","created":{"date-parts":[[2023,9,9]],"date-time":"2023-09-09T09:33:18Z","timestamp":1694251998000},"page":"1-26","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":20,"title":["MaGNAS: A Mapping-Aware Graph Neural Architecture Search Framework for Heterogeneous MPSoC Deployment"],"prefix":"10.1145","volume":"22","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0828-949X","authenticated-orcid":false,"given":"Mohanad","family":"Odema","sequence":"first","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1885-6080","authenticated-orcid":false,"given":"Halima","family":"Bouzidi","sequence":"additional","affiliation":[{"name":"Universit\u00e9 Polytechnique Hauts-de-France, France"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7490-5350","authenticated-orcid":false,"given":"Hamza","family":"Ouarnoughi","sequence":"additional","affiliation":[{"name":"Universit\u00e9 Polytechnique Hauts-de-France, France"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7550-484X","authenticated-orcid":false,"given":"Smail","family":"Niar","sequence":"additional","affiliation":[{"name":"Universit\u00e9 Polytechnique Hauts-de-France, France"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5390-0497","authenticated-orcid":false,"given":"Mohammad Abdullah","family":"Al Faruque","sequence":"additional","affiliation":[{"name":"University of California Irvine, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2023,9,9]]},"reference":[{"key":"e_1_3_1_2_2","unstructured":"[n.d.]. NVIDIA Jetson AGX Xavier Delivers 32 TeraOps for New Era of AI in Robotics. https:\/\/developer.nvidia.com\/blog\/nvidia-jetson-agx-xavier-32-teraops-ai-robotics\/"},{"key":"e_1_3_1_3_2","unstructured":"[n.d.]. Power management and clock frequency scaling. https:\/\/docs.nvidia.com\/jetson\/archives\/r34.1\/DeveloperGuide\/text\/SD\/PlatformPowerAndPerformance.html"},{"key":"e_1_3_1_4_2","first-page":"1","volume-title":"2020 57th ACM\/IEEE Design Automation Conference (DAC\u201920)","author":"Auten Adam","year":"2020","unstructured":"Adam Auten, Matthew Tomei, and Rakesh Kumar. 2020. Hardware acceleration of graph neural networks. In 2020 57th ACM\/IEEE Design Automation Conference (DAC\u201920). IEEE, 1\u20136."},{"key":"e_1_3_1_5_2","volume-title":"Design, Automation & Test in Europe Conference & Exhibition (DATE\u201923)","author":"Bouzidi Halima","year":"2023","unstructured":"Halima Bouzidi, Mohanad Odema, Hamza Ouarnoughi, Smail Niar, and Mohammad Abdullah Al Faruque. 2023. HADAS: Hardware-aware dynamic neural architecture search for edge performance scaling. In Design, Automation & Test in Europe Conference & Exhibition (DATE\u201923)."},{"key":"e_1_3_1_6_2","volume-title":"Proceedings of the 60th ACM\/IEEE Design Automation Conference (DAC)","author":"Bouzidi Halima","year":"2023","unstructured":"Halima Bouzidi, Mohanad Odema, Hamza Ouarnoughi, Smail Niar, and Mohammad Abdullah Al Faruque. 2023. Map-and-conquer: Energy-efficient mapping of dynamic neural nets onto heterogeneous MPSoCs. In Proceedings of the 60th ACM\/IEEE Design Automation Conference (DAC)."},{"key":"e_1_3_1_7_2","volume-title":"International Conference on Learning Representations (ICLR\u201919)","author":"Cai Han","year":"2019","unstructured":"Han Cai et\u00a0al. 2019. Once-for-all: Train one network and specialize it for efficient deployment. In International Conference on Learning Representations (ICLR\u201919)."},{"key":"e_1_3_1_8_2","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18074.2021.9586298"},{"key":"e_1_3_1_9_2","article-title":"A survey on graph neural networks and graph transformers in computer vision: A task-oriented perspective","author":"Chen Chaoqi","year":"2022","unstructured":"Chaoqi Chen and ohers. 2022. A survey on graph neural networks and graph transformers in computer vision: A task-oriented perspective. arXiv preprint arXiv:2209.13232 (2022).","journal-title":"arXiv preprint arXiv:2209.13232"},{"key":"e_1_3_1_10_2","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001177"},{"key":"e_1_3_1_11_2","volume-title":"Proceedings of the 59th ACM\/IEEE Design Automation Conference (DAC\u201922)","author":"Dagli Ismet","year":"2022","unstructured":"Ismet Dagli et\u00a0al. 2022. AxoNN: Energy-aware execution of neural network inference on multi-accelerator heterogeneous SoCs. In Proceedings of the 59th ACM\/IEEE Design Automation Conference (DAC\u201922)."},{"key":"e_1_3_1_12_2","doi-asserted-by":"crossref","first-page":"238","DOI":"10.23919\/DATE54114.2022.9774574","volume-title":"2022 Design, Automation & Test in Europe Conference & Exhibition (DATE\u201922)","author":"Fasfous Nael","year":"2022","unstructured":"Nael Fasfous et\u00a0al. 2022. Anaconga: Analytical hw-cnn co-design using nested genetic algorithms. In 2022 Design, Automation & Test in Europe Conference & Exhibition (DATE\u201922). IEEE, 238\u2013243."},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1145\/3289602.3293906"},{"key":"e_1_3_1_14_2","first-page":"1403","volume-title":"IJCAI","author":"Gao Yang","year":"2020","unstructured":"Yang Gao et\u00a0al. 2020. Graph neural architecture search.. In IJCAI, Vol. 20. 1403\u20131409."},{"key":"e_1_3_1_15_2","first-page":"1403","volume-title":"Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI-20","author":"Gao Yang","year":"2020","unstructured":"Yang Gao, Hong Yang, Peng Zhang, Chuan Zhou, and Yue Hu. 2020. Graph neural architecture search. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI-20. 1403\u20131409."},{"key":"e_1_3_1_16_2","doi-asserted-by":"crossref","first-page":"571","DOI":"10.1109\/IPDPS53621.2022.00062","volume-title":"2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS\u201922)","author":"Garg Raveesh","year":"2022","unstructured":"Raveesh Garg et\u00a0al. 2022. Understanding the design-space of sparse\/dense multiphase GNN dataflows on spatial accelerators. In 2022 IEEE International Parallel and Distributed Processing Symposium (IPDPS\u201922). IEEE, 571\u2013582."},{"key":"e_1_3_1_17_2","article-title":"Inductive representation learning on large graphs","volume":"30","author":"Hamilton Will","year":"2017","unstructured":"Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. Advances in Neural Information Processing Systems 30 (2017).","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_1_18_2","volume-title":"Advances in Neural Information Processing Systems","author":"Han Kai","year":"2022","unstructured":"Kai Han, Yunhe Wang, Jianyuan Guo, Yehui Tang, and Enhua Wu. 2022. Vision GNN: An image is worth graph of nodes. In Advances in Neural Information Processing Systems."},{"key":"e_1_3_1_19_2","article-title":"Energy-aware scenario-based mapping of deep learning applications onto heterogeneous processors under real-time constraints","author":"Kim Jangryul","year":"2022","unstructured":"Jangryul Kim and Soonhoi Ha. 2022. Energy-aware scenario-based mapping of deep learning applications onto heterogeneous processors under real-time constraints. IEEE Trans. Comput. (2022).","journal-title":"IEEE Trans. Comput."},{"key":"e_1_3_1_20_2","article-title":"GRIP: A graph neural network accelerator architecture","author":"Kiningham Kevin","year":"2022","unstructured":"Kevin Kiningham et\u00a0al. 2022. GRIP: A graph neural network accelerator architecture. IEEE Trans. Comput. (2022).","journal-title":"IEEE Trans. Comput."},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.1145\/3352460.3358252"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2020.2985963"},{"key":"e_1_3_1_23_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00479"},{"key":"e_1_3_1_24_2","first-page":"9267","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision","author":"Li Guohao","year":"2019","unstructured":"Guohao Li, Matthias Muller, Ali Thabet, and Bernard Ghanem. 2019. Deepgcns: Can gcns go as deep as cnns?. In Proceedings of the IEEE\/CVF International Conference on Computer Vision. 9267\u20139276."},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1145\/3173162.3173191"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2022.3141044"},{"key":"e_1_3_1_27_2","volume-title":"31st Euromicro Conference on Real-Time Systems (ECRTS\u201919)","volume":"23","author":"Pujol Roger","year":"2019","unstructured":"Roger Pujol, Hamid Tabani, Leonidas Kosmidis, Enrico Mezzetti, Jaume Abella Ferrer, and Francisco J Cazorla. 2019. Generating and exploiting deep learning variants to increase heterogeneous resource utilization in the nvidia xavier. In 31st Euromicro Conference on Real-Time Systems (ECRTS\u201919), Vol. 23."},{"issue":"1","key":"e_1_3_1_28_2","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/TEVC.2020.3013290","article-title":"A survey on the hypervolume indicator in evolutionary multiobjective optimization","volume":"25","author":"Shang Ke","year":"2020","unstructured":"Ke Shang, Hisao Ishibuchi, Linjun He, and Lie Meng Pang. 2020. A survey on the hypervolume indicator in evolutionary multiobjective optimization. IEEE Transactions on Evolutionary Computation 25, 1 (2020), 1\u201320.","journal-title":"IEEE Transactions on Evolutionary Computation"},{"key":"e_1_3_1_29_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2022.108752"},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18074.2021.9586122"},{"key":"e_1_3_1_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2020.2975764"},{"key":"e_1_3_1_32_2","doi-asserted-by":"publisher","DOI":"10.1145\/3326362"},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2020.2978386"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.330"},{"key":"e_1_3_1_35_2","volume-title":"International Conference on Learning Representations","author":"Xu Keyulu","year":"2019","unstructured":"Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2019. How powerful are graph neural networks?. In International Conference on Learning Representations."},{"key":"e_1_3_1_36_2","doi-asserted-by":"crossref","first-page":"1556","DOI":"10.23919\/DATE48585.2020.9116235","volume-title":"2020 Design, Automation & Test in Europe Conference & Exhibition (DATE)","author":"Xun Lei","year":"2020","unstructured":"Lei Xun et\u00a0al. 2020. Optimising resource management for embedded machine learning. In 2020 Design, Automation & Test in Europe Conference & Exhibition (DATE). IEEE, 1556\u20131561."},{"key":"e_1_3_1_37_2","first-page":"15","volume-title":"2020 IEEE International Symposium on High Performance Computer Architecture (HPCA\u201920)","author":"Yan Mingyu","year":"2020","unstructured":"Mingyu Yan, Lei Deng, Xing Hu, Ling Liang, Yujing Feng, Xiaochun Ye, Zhimin Zhang, Dongrui Fan, and Yuan Xie. 2020. Hygcn: A gcn accelerator with hybrid architecture. In 2020 IEEE International Symposium on High Performance Computer Architecture (HPCA\u201920). IEEE, 15\u201329."},{"key":"e_1_3_1_38_2","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v32i1.12328"},{"key":"e_1_3_1_39_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01246-5_41"},{"key":"e_1_3_1_40_2","first-page":"460","volume-title":"2022 IEEE International Symposium on High-Performance Computer Architecture (HPCA\u201922)","author":"You Haoran","year":"2022","unstructured":"Haoran You et\u00a0al. 2022. Gcod: Graph convolutional network acceleration via dedicated algorithm and accelerator co-design. In 2022 IEEE International Symposium on High-Performance Computer Architecture (HPCA\u201922). IEEE, 460\u2013474."},{"key":"e_1_3_1_41_2","first-page":"17009","article-title":"Design space for graph neural networks","volume":"33","author":"You Jiaxuan","year":"2020","unstructured":"Jiaxuan You, Zhitao Ying, and Jure Leskovec. 2020. Design space for graph neural networks. Advances in Neural Information Processing Systems 33 (2020), 17009\u201317021.","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_3_1_42_2","volume-title":"International Conference on Learning Representations","author":"Yu Jiahui","year":"2019","unstructured":"Jiahui Yu et\u00a0al. 2019. Slimmable neural networks. In International Conference on Learning Representations."},{"key":"e_1_3_1_43_2","volume-title":"Computer Vision\u2013ECCV 2020: 16th European Conference, Glasgow, UK, August 23\u201328, 2020, Proceedings, Part VII 16","author":"Yu Jiahui","year":"2020","unstructured":"Jiahui Yu et\u00a0al. 2020. Bignas: Scaling up neural architecture search with big single-stage models. In Computer Vision\u2013ECCV 2020: 16th European Conference, Glasgow, UK, August 23\u201328, 2020, Proceedings, Part VII 16. Springer."},{"issue":"7","key":"e_1_3_1_44_2","first-page":"7941","article-title":"Scene-graph augmented data-driven risk assessment of autonomous vehicle decisions","volume":"23","author":"Yu Shih-Yuan","year":"2021","unstructured":"Shih-Yuan Yu, Arnav Vaibhav Malawade, Deepan Muthirayan, Pramod P. Khargonekar, and Mohammad Abdullah Al Faruque. 2021. Scene-graph augmented data-driven risk assessment of autonomous vehicle decisions. IEEE Transactions on Intelligent Transportation Systems 23, 7 (2021), 7941\u20137951.","journal-title":"IEEE Transactions on Intelligent Transportation Systems"},{"key":"e_1_3_1_45_2","first-page":"11","volume-title":"2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics (HiPC\u201922)","author":"Zhang Bingyi","year":"2022","unstructured":"Bingyi Zhang et\u00a0al. 2022. Low-latency mini-batch GNN inference on CPU-FPGA heterogeneous platform. In 2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics (HiPC\u201922). 11\u201321."},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD51958.2021.9643549"},{"key":"e_1_3_1_47_2","volume-title":"Proceedings of the 60th ACM\/IEEE Design Automation Conference (DAC\u201923)","author":"Zhou Ao","year":"2023","unstructured":"Ao Zhou et\u00a0al. 2023. Hardware-aware graph neural network automated design for edge computing platforms. In Proceedings of the 60th ACM\/IEEE Design Automation Conference (DAC\u201923)."},{"key":"e_1_3_1_48_2","article-title":"Auto-gnn: Neural architecture search of graph neural networks","author":"Zhou Kaixiong","year":"2022","unstructured":"Kaixiong Zhou et\u00a0al. 2022. Auto-gnn: Neural architecture search of graph neural networks. Frontiers in Big Data (2022).","journal-title":"Frontiers in Big Data"}],"container-title":["ACM Transactions on Embedded Computing Systems"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3609386","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3609386","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,17]],"date-time":"2025-06-17T12:46:23Z","timestamp":1750164383000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3609386"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,9,9]]},"references-count":47,"journal-issue":{"issue":"5s","published-print":{"date-parts":[[2023,10,31]]}},"alternative-id":["10.1145\/3609386"],"URL":"https:\/\/doi.org\/10.1145\/3609386","relation":{},"ISSN":["1539-9087","1558-3465"],"issn-type":[{"value":"1539-9087","type":"print"},{"value":"1558-3465","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,9,9]]},"assertion":[{"value":"2023-03-23","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-06-30","order":1,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-09-09","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}