Unsupervised Outlier Detection based on Random Projection Outlyingness with Local Score Weighting

This paper proposes an enhanced model of Random Projection Outlyingness (RPO) for unsupervised outlier detection. When datasets have multiple modalities, the RPOs frequent detection errors. The proposed deals with this problem via clustering and a local score weighting. experimental results demonstrate that outperforms RPO is comparable other existing models on benchmark datasets, in terms Area Under Curves (AUCs) Receiver Operating Characteristic (ROC).