Automating Outlier Detection via Meta-Learning
This addresses the challenge of model selection in unsupervised outlier detection, which has been a 'black art' due to lack of labels and universal objectives, offering a principled solution for practitioners in data analysis and anomaly detection.
The paper tackles the problem of automatically selecting an effective outlier detection model and hyperparameters for new datasets without labeled data, by developing MetaOD, a meta-learning approach that uses past performance data and specialized meta-features. The result shows MetaOD significantly outperforms popular and state-of-the-art detectors, being extremely fast in model selection.
Given an unsupervised outlier detection (OD) task on a new dataset, how can we automatically select a good outlier detection method and its hyperparameter(s) (collectively called a model)? Thus far, model selection for OD has been a "black art"; as any model evaluation is infeasible due to the lack of (i) hold-out data with labels, and (ii) a universal objective function. In this work, we develop the first principled data-driven approach to model selection for OD, called MetaOD, based on meta-learning. MetaOD capitalizes on the past performances of a large body of detection models on existing outlier detection benchmark datasets, and carries over this prior experience to automatically select an effective model to be employed on a new dataset without using any labels. To capture task similarity, we introduce specialized meta-features that quantify outlying characteristics of a dataset. Through comprehensive experiments, we show the effectiveness of MetaOD in selecting a detection model that significantly outperforms the most popular outlier detectors (e.g., LOF and iForest) as well as various state-of-the-art unsupervised meta-learners while being extremely fast. To foster reproducibility and further research on this new problem, we open-source our entire meta-learning system, benchmark environment, and testbed datasets.