ResGCN: attention-based deep residual modeling for anomaly detection on attributed networks

Abstract Effectively detecting anomalous nodes in attributed networks is crucial for the success of many real-world applications such as fraud and intrusion detection. Existing approaches have difficulties with three major issues: sparsity nonlinearity capturing, residual modeling, network smoothing. We propose Residual Graph Convolutional Network (ResGCN), an attention-based deep modeling approach that can tackle these GCN allows to capture nonlinearity, utilizing a neural direct ing from input, residual-based attention mechanism reduces adverse effect prevents over-smoothing. Extensive experiments on several demonstrate effectiveness ResGCN anomalies.