Kernel Random Projection Depth for Outlier Detection
This is an incremental improvement for outlier detection in complex data structures.
The paper tackles the problem of outlier detection in data with multiple modalities and non-convexity by extending Random Projection Depth (RPD) using kernel methods, resulting in performance that outperforms RPD and is comparable to other models on benchmark datasets as measured by AUCs of ROC.
This paper proposes an extension of Random Projection Depth (RPD) to cope with multiple modalities and non-convexity on data clouds. In the framework of the proposed method, the RPD is computed in a reproducing kernel Hilbert space. With the help of kernel principal component analysis, we expect that the proposed method can cope with the above multiple modalities and non-convexity. The experimental results demonstrate that the proposed method outperforms RPD and is comparable to other existing detection models on benchmark datasets regarding Area Under the Curves (AUCs) of Receiver Operating Characteristic (ROC).