Performance of Graph Neural Networks for Point Cloud Applications

被引:1
|
作者
Parikh, Dhruv [1 ]
Zhang, Bingyi [1 ]
Kannan, Rajgopal [2 ]
Prasanna, Viktor [1 ]
Busart, Carl [2 ]
机构
[1] Univ Southern Calif, Los Angeles, CA 90007 USA
[2] DEVCOM US Army Res Lab, Adelphi, MD USA
来源
2023 IEEE HIGH PERFORMANCE EXTREME COMPUTING CONFERENCE, HPEC | 2023年
关键词
Graph neural network; point cloud; k-nearest neighbors; dynamic graph construction; performance profiling;
D O I
10.1109/HPEC58863.2023.10363595
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
Graph Neural Networks (GNNs) have gained significant momentum recently due to their capability to learn on unstructured graph data. Dynamic GNNs (DGNNs) are the current state-of-the-art for point cloud applications; such applications (viz. autonomous driving) require real-time processing at the edge with tight latency and memory constraints. Conducting performance analysis on such DGNNs, thus, becomes a crucial task to evaluate network suitability. This paper presents a profiling analysis of EdgeConv-based DGNNs applied to point cloud inputs. We assess their inference performance in terms of end-to-end latency and memory consumption on state-of-the-art CPU and GPU platforms. The EdgeConv layer has two stages: (1) dynamic graph generation using k-Nearest Neighbors (kNN) and, (2) node feature updation. The addition of dynamic graph generation via kNN in each (EdgeConv) layer enhances network performance compared to networks that work with the same static graph in each layer; such performance enhancement comes, however, at the added computational cost associated with the dynamic graph generation stage (via kNN algorithm). Understanding its costs is essential for identifying the performance bottleneck and exploring potential avenues for hardware acceleration. To this end, this paper aims to shed light on the performance characteristics of EdgeConv-based DGNNs for point cloud inputs. Our performance analysis on a state-of-the-art EdgeConv network for classification shows that the dynamic graph construction via kNN takes up upwards of 95% of network latency on the GPU and almost 90% on the CPU. Moreover, we propose a quasi-Dynamic Graph Neural Network (qDGNN) that halts dynamic graph updates after a specific depth within the network to significantly reduce the latency on both CPU and GPU whilst matching the original networks inference accuracy.
引用
收藏
页数:7
相关论文
共 50 条
  • [41] Using three dimensional convolutional neural networks for denoising echosounder point cloud data
    Stephens, David
    Smith, Andrew
    Redfern, Thomas
    Talbot, Andrew
    Lessnoff, Andrew
    Dempsey, Kari
    APPLIED COMPUTING AND GEOSCIENCES, 2020, 5
  • [42] Artificial neural networks exploiting point cloud data for fragmented solid objects classification
    Baiocchi, A.
    Giagu, S.
    Napoli, C.
    Serra, M.
    Nardelli, P.
    Valleriani, M.
    MACHINE LEARNING-SCIENCE AND TECHNOLOGY, 2023, 4 (04):
  • [43] Using 3D point cloud and graph-based neural networks to improve the estimation of pulmonary function tests from chest CT
    Jia, Jingnan
    Yu, Bo
    Mody, Prerak
    Ninaber, Maarten K.
    Schouffoer, Anne A.
    de Vries-Bouwstra, Jeska K.
    Kroft, Lucia J.M.
    Staring, Marius
    Stoel, Berend C.
    Computers in Biology and Medicine, 2024, 182
  • [44] DeGNN: Improving Graph Neural Networks with Graph Decomposition
    Miao, Xupeng
    Gurel, Nezihe Merve
    Zhang, Wentao
    Han, Zhichao
    Li, Bo
    Min, Wei
    Rao, Susie Xi
    Ren, Hansheng
    Shan, Yinan
    Shao, Yingxia
    Wang, Yujie
    Wu, Fan
    Xue, Hui
    Yang, Yaming
    Zhang, Zitao
    Zhao, Yang
    Zhang, Shuai
    Wang, Yujing
    Cui, Bin
    Zhang, Ce
    KDD '21: PROCEEDINGS OF THE 27TH ACM SIGKDD CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING, 2021, : 1223 - 1233
  • [45] GGM-Net: Graph Geometric Moments Convolution Neural Network for Point Cloud Shape Classification
    Li, Dilong
    Shen, Xin
    Yu, Yongtao
    Guan, Haiyan
    Wang, Hanyun
    Li, Deren
    IEEE ACCESS, 2020, 8 : 124989 - 124998
  • [46] POINT CLOUD ATTRIBUTE INPAINTING IN GRAPH SPECTRAL DOMAIN
    He, Ju
    Fu, Zeqing
    Hu, Wei
    Guo, Zongming
    2019 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2019, : 4385 - 4389
  • [47] PSR-GAT: Arbitrary point cloud super-resolution using graph attention networks
    Fan Zhong
    Zhengyao Bai
    Multimedia Tools and Applications, 2024, 83 : 26213 - 26232
  • [48] Network Design Through Graph Neural Networks: Identifying Challenges and Improving Performance
    Loveland, Donald
    Caceres, Rajmonda
    COMPLEX NETWORKS & THEIR APPLICATIONS XII, VOL 1, COMPLEX NETWORKS 2023, 2024, 1141 : 3 - 15
  • [49] Graph-Guided Deformation for Point Cloud Completion
    Shi, Jieqi
    Xu, Lingyun
    Heng, Liang
    Shen, Shaojie
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2021, 6 (04): : 7081 - 7088
  • [50] Differential Graph Convolution Network for point cloud understanding
    Bai, Yun
    Li, Guanlin
    Yang, Chaozhi
    Li, Yachuan
    Xiao, Qian
    Li, Zongmin
    NEUROCOMPUTING, 2024, 597