Kernel Random Projection Depth for Outlier Detection

About

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).

Akira Tamamori · 2023

Related benchmarks

Task	Dataset	Result
Anomaly Detection	WBC	ROCAUC0.928	151
Anomaly Detection	Glass	AUC-ROC0.72	55
Outlier Detection	Arrhythmia	AUC-ROC76.5	35
Outlier Detection	Optdigits	AUC-ROC0.926	33
Anomaly Detection	Vowels	AUC-ROC0.784	30
Outlier Detection	cardio	AUC-ROC97.2	24
Outlier Detection	ionosphere	AUC0.934	24
Outlier Detection	Vertebral	ROC AUC65.1	13
Outlier Detection	Letter	ROC AUC87.3	13
Outlier Detection	pima	ROC AUC0.691	13