EAGLE: Contrastive Learning for Efficient Graph Anomaly Detection

Graph anomaly detection is a popular and vital task in various real-world scenarios, which has been studied for several decades. Recently, many studies extending deep learning-based methods have shown preferable performance on graph anomaly detection. However, existing methods lack efficiency that is definitely necessary for embedded devices. Toward this end, we propose an Efficient Anomaly detection model on heterogeneous Graphs via contrastive LEarning (EAGLE) by contrasting abnormal nodes with normal ones in terms of their distances to the local context. The proposed method first samples instance pairs on meta-path level for contrastive learning. Then, a Graph AutoEncoder-based model is applied to learn informative node embeddings in an unsupervised way, which will be further combined with the discriminator to predict the anomaly scores of nodes. Experimental results show that EAGLE outperforms the state-of-the-art methods on three heterogeneous network datasets.

Authors

• Jing Ren College of Computer Science and Technology, National University of Defense Technology
• Mingliang Hou Jinan University TAL Education Group
• Zhixuan Liu Soochow University
• Xiaomei Bai

Keywords

• Anomaly detection
• Task analysis
• Representation learning
• Intelligent systems
• Semantics
• Data models
• Computational modeling
• Self-supervised Learning
• Semantic
• Deep Learning
• Graph Data
• Heterogeneous Network
• Ground Truth Labels
• Human Experts
• Recommender Systems
• Graph Convolutional Network
• Graph Neural Networks
• Heterogeneous Datasets
• Target Node
• Positive Instances
• Node Score
• Negative Instances
• Node Embeddings
• Heterogeneous Graph
• Anomaly Score
• Deep Autoencoder
• Normal Nodes
• Positive Samples
• Negative Samples
• Nodes In The Graph
• Node Representations
• Data Augmentation
• Discrimination Scores
• Average Pooling
• Embedding Dimension