Robust Multi-Kernel Nearest Neighborhood for Outlier Detection

• Published in: IEEE transactions on knowledge and data engineering Vol. 36; no. 8; pp. 4220 - 4231
• Main Authors: Wang, Xinye, Duan, Lei, Yu, Zhenyang, He, Chengxin, Bao, Zhifeng
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.08.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Anomaly detection; Data analysis; Effectiveness; Feature extraction; Games; Graph theory; Kernel; Kernel functions; multi-kernel learning; nearest neighborhood; Outlier detection; Outliers (statistics); Polynomials; Robustness; Support vector machines; Task analysis; Training; weighted digraph
• ISSN: 1041-4347; 1558-2191
• DOI: 10.1109/TKDE.2024.3364179

Outlier detection methods based on distance measure have been used in numerous applications due to their effectiveness and interpretability. However, distances among instances heavily depend on the feature space in which they reside. For an outlier, distances from it to the normal instances may be extremely close in one feature space, failing to separate them from each other, while this situation is reversed in another space. Meanwhile, the distance measure is sensitive to a few "marginal instances" (i.e., normal instances located very close to outliers in the feature space) during the estimation of whether a test instance is an outlier or not. In this article, we propose a r obust m ulti- k ernel nearest n eighborhood (RMKN) method for outlier detection. Specifically, in the training phase, we only consider normal instances and transform them into a Polynomial kernel function weighted digraph to capture their geometric relationships in the original feature space. Then, we develop an objective function based on the weighted digraph to find a latent feature space via multi-kernel learning such that distances among normal instances in this latent feature space are as close as possible while preserving their original distributions. In the detecting phase, we design an outlying score based on the two-stage multi-kernel <inline-formula><tex-math notation="LaTeX">k</tex-math> <mml:math><mml:mi>k</mml:mi></mml:math><inline-graphic xlink:href="wang-ieq1-3364179.gif"/> </inline-formula>-nearest nearest neighbors to detect outliers. Extensive experiments with ten datasets show that RMKN is effective and robust.