Poisson kernel: avoiding self-smoothing in graph convolutional networks

Graph convolutional network is now an effective tool to deal with non-Euclidean data, such as social behavior analysis, molecular structure analysis, and skeleton-based action recognition. Graph convolutional kernel is one of the most significant factors in graph convolutional networks to extract no...

Full description

Saved in:
Bibliographic Details
Main Authors: Yang, Ziqing, Han, Shoudong, Zhao, Jun
Other Authors: School of Computer Science and Engineering
Format: Article
Language:English
Published: 2022
Subjects:
Online Access:https://hdl.handle.net/10356/162583
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-162583
record_format dspace
spelling sg-ntu-dr.10356-1625832022-10-31T05:41:01Z Poisson kernel: avoiding self-smoothing in graph convolutional networks Yang, Ziqing Han, Shoudong Zhao, Jun School of Computer Science and Engineering Engineering::Computer science and engineering Graph Convolutional Kernel Graph Convolutional Network Graph convolutional network is now an effective tool to deal with non-Euclidean data, such as social behavior analysis, molecular structure analysis, and skeleton-based action recognition. Graph convolutional kernel is one of the most significant factors in graph convolutional networks to extract nodes’ feature, and some variants of it have achieved highly satisfactory performance theoretically and experimentally. However, there was limited research about how exactly different graph structures influence the performance of these kernels. Some existing methods used an adaptive convolutional kernel to deal with a given graph structure, which still not explore the internal reasons. In this paper, we start from theoretical analysis of the spectral graph and study the properties of existing graph convolutional kernels, revealing the self-smoothing phenomenon and its effect in specific structured graphs. After that, we propose the Poisson kernel that can avoid self-smoothing without training any adaptive kernel. Experimental results demonstrate that our Poisson kernel not only works well on the benchmark datasets where state-of-the-art methods work fine, but also is evidently superior to them in synthetic datasets. This work was supported by the National Natural Science Foundation of China under Grant No. 61105006; Open Fund of Key Laboratory of Image Processing and Intelligent Control (Huazhong University of Science and Technology), Ministry of Education under Grant No. IPIC2019-01; and the China Scholarship Council under Grant No. 201906165066. 2022-10-31T05:41:01Z 2022-10-31T05:41:01Z 2022 Journal Article Yang, Z., Han, S. & Zhao, J. (2022). Poisson kernel: avoiding self-smoothing in graph convolutional networks. Pattern Recognition, 124, 108443-. https://dx.doi.org/10.1016/j.patcog.2021.108443 0031-3203 https://hdl.handle.net/10356/162583 10.1016/j.patcog.2021.108443 2-s2.0-85120421532 124 108443 en Pattern Recognition © 2021 Elsevier Ltd. All rights reserved.
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Computer science and engineering
Graph Convolutional Kernel
Graph Convolutional Network
spellingShingle Engineering::Computer science and engineering
Graph Convolutional Kernel
Graph Convolutional Network
Yang, Ziqing
Han, Shoudong
Zhao, Jun
Poisson kernel: avoiding self-smoothing in graph convolutional networks
description Graph convolutional network is now an effective tool to deal with non-Euclidean data, such as social behavior analysis, molecular structure analysis, and skeleton-based action recognition. Graph convolutional kernel is one of the most significant factors in graph convolutional networks to extract nodes’ feature, and some variants of it have achieved highly satisfactory performance theoretically and experimentally. However, there was limited research about how exactly different graph structures influence the performance of these kernels. Some existing methods used an adaptive convolutional kernel to deal with a given graph structure, which still not explore the internal reasons. In this paper, we start from theoretical analysis of the spectral graph and study the properties of existing graph convolutional kernels, revealing the self-smoothing phenomenon and its effect in specific structured graphs. After that, we propose the Poisson kernel that can avoid self-smoothing without training any adaptive kernel. Experimental results demonstrate that our Poisson kernel not only works well on the benchmark datasets where state-of-the-art methods work fine, but also is evidently superior to them in synthetic datasets.
author2 School of Computer Science and Engineering
author_facet School of Computer Science and Engineering
Yang, Ziqing
Han, Shoudong
Zhao, Jun
format Article
author Yang, Ziqing
Han, Shoudong
Zhao, Jun
author_sort Yang, Ziqing
title Poisson kernel: avoiding self-smoothing in graph convolutional networks
title_short Poisson kernel: avoiding self-smoothing in graph convolutional networks
title_full Poisson kernel: avoiding self-smoothing in graph convolutional networks
title_fullStr Poisson kernel: avoiding self-smoothing in graph convolutional networks
title_full_unstemmed Poisson kernel: avoiding self-smoothing in graph convolutional networks
title_sort poisson kernel: avoiding self-smoothing in graph convolutional networks
publishDate 2022
url https://hdl.handle.net/10356/162583
_version_ 1749179176415395840