Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Incomplete Graph Representation and Learning via Partial Graph Neural Networks

About

Graph Neural Networks (GNNs) are gaining increasing attention on graph data learning tasks in recent years. However, in many applications, graph may be coming in an incomplete form where attributes of graph nodes are partially unknown/missing. Existing GNNs are generally designed on complete graphs which can not deal with attribute-incomplete graph data directly. To address this problem, we develop a novel partial aggregation based GNNs, named Partial Graph Neural Networks (PaGNNs), for attribute-incomplete graph representation and learning. Our work is motivated by the observation that the neighborhood aggregation function in standard GNNs can be equivalently viewed as the neighborhood reconstruction formulation. Based on it, we define two novel partial aggregation (reconstruction) functions on incomplete graph and derive PaGNNs for incomplete graph data learning. Extensive experiments on several datasets demonstrate the effectiveness and efficiency of the proposed PaGNNs.

Bo Jiang, Ziyan Zhang• 2020

Related benchmarks

TaskDatasetResultRank
Node ClassificationReddit (test)
Accuracy90.57
134
Node ClassificationCiteSeer Uniform Missing (test)
Accuracy49.36
16
Inductive Node ClassificationFlickr (test)
Accuracy49.46
14
Node ClassificationPubMed Structural Missing (test)
Accuracy64.74
14
Node ClassificationPubMed Uniform Missing (test)
Accuracy69.44
14
Node ClassificationOGBN-Arxiv uniform missing (test)
Accuracy62.49
13
Node ClassificationOGBN-Arxiv structural missing (test)
Accuracy60.7
13
Transductive Node ClassificationSailing Natural missing features (test)
Accuracy67.47
8
Transductive Node ClassificationCora Structural missing features (test)
Accuracy55.26
8
Transductive Node ClassificationCora Uniform missing features (test)
Accuracy65.3
8
Showing 10 of 12 rows

Other info

Follow for update