Abstract:Graph Neural Networks (GNNs) have shown remarkable merit in performing various learning-based tasks in complex networks. The superior performance of GNNs often correlates with the availability and quality of node-level features in the input networks. However, for many network applications, such node-level information may be missing or unreliable, thereby limiting the applicability and efficacy of GNNs. To address this limitation, we present a novel approach denoted as Ego-centric Spectral subGraph Embedding Augmentation (ESGEA), which aims to enhance and design node features, particularly in scenarios where information is lacking. Our method leverages the topological structure of the local subgraph to create topology-aware node features. The subgraph features are generated using an efficient spectral graph embedding technique, and they serve as node features that capture the local topological organization of the network. The explicit node features, if present, are then enhanced with the subgraph embeddings in order to improve the overall performance. ESGEA is compatible with any GNN-based architecture and is effective even in the absence of node features. We evaluate the proposed method in a social network graph classification task where node attributes are unavailable, as well as in a node classification task where node features are corrupted or even absent. The evaluation results on seven datasets and eight baseline models indicate up to a 10% improvement in AUC and a 7% improvement in accuracy for graph and node classification tasks, respectively.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper attempts to address a key issue faced by Graph Neural Networks (GNNs) when handling complex network tasks: the absence or unreliability of node features. Specifically, the performance of GNNs is often closely tied to the quality and availability of node-level features in the input network. However, in many network applications, this node-level information may be missing or unreliable, limiting the applicability and effectiveness of GNNs. To tackle this challenge, the authors propose a novel method called Ego-centric Spectral subGraph Embedding Augmentation (ESGEA). This method aims to generate or enhance node features by leveraging the topological structure of local subgraphs, particularly in situations where information is lacking. ESGEA uses efficient spectral graph embedding techniques to generate subgraph features and uses these features as node features to capture the local topological organization of the network. If explicit node features are present, these features are further enhanced through subgraph embeddings, thereby improving overall performance. ### Main Contributions 1. **Proposed a new topological feature enhancement method**: ESGEA can design new node features or enhance existing but damaged or missing node features. 2. **Introduced a flexible framework**: This framework combines spectral graph embedding and GNNs, providing a flexible graph representation pipeline for different applications. 3. **Extensive experimental evaluation**: Conducted extensive experiments on graph classification and node classification tasks in the absence or damage of node features, validating the effectiveness of the method. Experimental results show that ESGEA achieved significant performance improvements on multiple datasets and baseline models, such as a 10% increase in AUC for graph classification tasks and a 7% increase in accuracy for node classification tasks. ### Experimental Setup - **Datasets**: The authors conducted experiments on five graph classification datasets without node features (GitHub Stargazers, Reddit threads, Reddit Binary, Deezer Egos, and Twitch Egos) and one node classification dataset with damaged node features (Facebook). - **Baseline models**: Including classical GNN models such as GCN, GraphSAGE, RGGCN, and more advanced models like UniMP and k-GNN. - **Evaluation metrics**: Mainly used AUC and accuracy to evaluate model performance. ### Conclusion The ESGEA method can significantly improve the performance of GNNs in the absence or unreliability of node features. By leveraging the topological structure of local subgraphs to generate or enhance node features, this method provides an effective solution for graph learning. Future research directions may include further optimizing subgraph embedding techniques and extending to more application scenarios.

Enhanced Graph Neural Networks with Ego-Centric Spectral Subgraph Embeddings Augmentation

NGAT: Attention in Breadth and Depth Exploration for Semi-Supervised Graph Representation Learning

Enhancing Graph Neural Networks by a High-quality Aggregation of Beneficial Information

Enhancing Graph Neural Networks with Edge Features Through Sequential Representation

Ego-GNNs: Exploiting Ego Structures in Graph Neural Networks

Self-Enhanced GNN: Improving Graph Neural Networks Using Model Outputs

Automorphic Equivalence-aware Graph Neural Network

Ergcn: Data Enhancement Based Robust Graph Convolutional Network for Node Classification

From Local to Global: Spectral-Inspired Graph Neural Networks

On the Power of Graph Neural Networks and Feature Augmentation Strategies to Classify Social Networks

Local Augmentation for Graph Neural Networks

GAIN: Graph Attention & Interaction Network for Inductive Semi-Supervised Learning over Large-scale Graphs

Graph Classification via Discriminative Edge Feature Learning

Feature Expansion for Graph Neural Networks

ECGN: A Cluster-Aware Approach to Graph Neural Networks for Imbalanced Classification

Graph Data Augmentation for Node Classification

Self-attention empowered graph convolutional network for structure learning and node embedding

A data-centric framework of improving graph neural networks for knowledge graph embedding

From Stars to Subgraphs: Uplifting Any GNN with Local Structure Awareness

Efficient Topology-aware Data Augmentation for High-Degree Graph Neural Networks