TY - JOUR
T1 - Simplicial Complex Neural Networks
AU - Wu, Hanrui
AU - Yip, Andy
AU - Long, Jinyi
AU - Zhang, Jia
AU - Ng, Michael K.
N1 - Funding information:
This work was supported in part by theNationalNatural Science Foundation of China under Grants 62206111, 62276115, and 62106084, in part by the Young Talent Support Project of Guangzhou Association for Science and Technology under Grant QT-2023-017, in part by Guangzhou Basic and Applied Basic Research Foundation under Grants 2023A04J1058 and 202201010498, in part by China Postdoctoral Science Foundation under Grants 2022M721343 and 2022M711331, in part by the National Natural Science Foundation of Guangdong, China under Grants 2019A1515012175 and 2022A1515010468, in part by the Outstanding Youth Project of Guangdong Natural Science Foundation of China under Grant 2021B1515020076, in part by Guangdong Provincial Key Laboratory of Traditional Chinese Medicine Informatization under Grant 2021B1212040007, in part by the Science and Technology Planning Project of Guangdong Province under Grant 2023A0505050092, and in part by the Fundamental Research Funds for Central Universities under Grant 21622326. The work of Michael K. Ng was supported in part by Hong Kong Research Grant Council GRF under Grants 17201020, 17300021, CRF C7004-21GF, and Joint NSFC-RGC N-HKU76921.
Publisher Copyright:
© 2023 IEEE.
PY - 2024/1
Y1 - 2024/1
N2 - Graph-structured data, where nodes exhibit either pair-wise or high-order relations, are ubiquitous and essential in graph learning. Despite the great achievement made by existing graph learning models, these models use the direct information (edges or hyperedges) from graphs and do not adopt the underlying indirect information (hidden pair-wise or high-order relations). To address this issue, in this paper, we propose a general framework named Simplicial Complex Neural (SCN) network, in which we construct a simplicial complex based on the direct and indirect graph information from a graph so that all information can be employed in the complex network learning. Specifically, we learn representations of simplices by aggregating and integrating information from all the simplices together via layer-by-layer simplicial complex propagation. In consequence, the representations of nodes, edges, and other high-order simplices are obtained simultaneously and can be used for learning purposes. By making use of block matrix properties, we derive the theoretical bound of the simplicial complex filter learnt by the propagation and establish the generalization error bound of the proposed simplicial complex network. We perform extensive experiments on node (0-simplex), edge (1-simplex), and triangle (2-simplex) classifications, and promising results demonstrate the performance of the proposed method is better than that of existing graph and hypergraph network approaches.
AB - Graph-structured data, where nodes exhibit either pair-wise or high-order relations, are ubiquitous and essential in graph learning. Despite the great achievement made by existing graph learning models, these models use the direct information (edges or hyperedges) from graphs and do not adopt the underlying indirect information (hidden pair-wise or high-order relations). To address this issue, in this paper, we propose a general framework named Simplicial Complex Neural (SCN) network, in which we construct a simplicial complex based on the direct and indirect graph information from a graph so that all information can be employed in the complex network learning. Specifically, we learn representations of simplices by aggregating and integrating information from all the simplices together via layer-by-layer simplicial complex propagation. In consequence, the representations of nodes, edges, and other high-order simplices are obtained simultaneously and can be used for learning purposes. By making use of block matrix properties, we derive the theoretical bound of the simplicial complex filter learnt by the propagation and establish the generalization error bound of the proposed simplicial complex network. We perform extensive experiments on node (0-simplex), edge (1-simplex), and triangle (2-simplex) classifications, and promising results demonstrate the performance of the proposed method is better than that of existing graph and hypergraph network approaches.
KW - Block matrices
KW - edge classification
KW - generalization error
KW - graph learning networks
KW - high-order simplex classification
KW - node classification
KW - simplicial complex
UR - http://www.scopus.com/inward/record.url?scp=85174828964&partnerID=8YFLogxK
U2 - 10.1109/TPAMI.2023.3323624
DO - 10.1109/TPAMI.2023.3323624
M3 - Journal article
C2 - 37831564
AN - SCOPUS:85174828964
SN - 0162-8828
VL - 46
SP - 561
EP - 575
JO - IEEE Transactions on Pattern Analysis and Machine Intelligence
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
IS - 1
ER -