Exploring High-Order Correlations for Industry Anomaly Detection

Anomaly detection aims to find the outlier noise data, which has attracted much attention in recent years. However, in most industrial cases, we can only obtain very few labeled anomalous data with abundant unlabeled data, which makes anomaly detection intractable. To tackle this issue, we propose a vertex-weighted hypergraph structure (VWHL) and a learning algorithm on it for anomaly detection. In this method, the correlation among data is formulated in a hypergraph structure, where each vertex denotes one sample, and the connections (hyperedges) on the hypergraph represent the relation among samples in feature space. We introduce vertex weights to investigate the importance of different samples, where each vertex is initialized with a weight corresponding to its similarity score and isolation score. Then, we learn the label projection matrix for anomaly detection and optimize vertex weights simultaneously. In this way, the vertices with high weights, which indicate data likely to be anomalies, play an important role during the learning process. We have evaluated our method on Industry Anomaly Detection dataset, Outlier Detection dataset (ODDS) and Software Defect Prediction (SDP) dataset, and experimental results show that our method achieves better performance when compared with other state-of-the-art methods.