Proceedings of the ACM Web Conference 2022 2022
DOI: 10.1145/3485447.3512185
|View full text |Cite
|
Sign up to set email alerts
|

Designing the Topology of Graph Neural Networks: A Novel Feature Fusion Perspective

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

1
13
0

Year Published

2022
2022
2025
2025

Publication Types

Select...
7

Relationship

1
6

Authors

Journals

citations
Cited by 16 publications
(14 citation statements)
references
References 16 publications
1
13
0
Order By: Relevance
“…We further adopt NAS methods for GNN and compare with the baseline and 1st solution coming from the industry. We choose the recent F 2 GNN (Wei et al, 2022 ) in our experiment, which searches for data-specific GNN topology. To compare fairly with GCN baselines, we fix the aggregation to GCN and search only the GNN topology, which we call F 2 GCN.…”
Section: Resultsmentioning
confidence: 99%
See 1 more Smart Citation
“…We further adopt NAS methods for GNN and compare with the baseline and 1st solution coming from the industry. We choose the recent F 2 GNN (Wei et al, 2022 ) in our experiment, which searches for data-specific GNN topology. To compare fairly with GCN baselines, we fix the aggregation to GCN and search only the GNN topology, which we call F 2 GCN.…”
Section: Resultsmentioning
confidence: 99%
“…DiffMG (Ding et al, 2021 ) proposed to use NAS to search data-specific meta-graphs in the heterogeneous graph, and PAS (Wei et al, 2021 ) is proposed to search data-specific pooling architectures for graph classification. The recently proposed F 2 GNN (Wei et al, 2022 ) method decouples the design of aggregation operations with architecture topology, which is not considered before.…”
Section: Introductionmentioning
confidence: 99%
“…According to the construction principles of graph neural modules, the search space can be divided into four categories: micro search space, macro search space, pooling methods, and hyperparameters. Numerous studies [13], [69] have demonstrated that automating the design of the best aggregation operation and network topology can improve model capacity. Search strategies include reinforcement learning, differentiable methods, evolutionary algorithms, and hybrid methods.…”
Section: Mtgrl and Gnasmentioning
confidence: 99%
“…Considering the GNN model with skip connections [13], [74], [75], the node representation h (k) at layer k is updated as:…”
Section: A Preliminary Knowledgementioning
confidence: 99%
See 1 more Smart Citation