Wisdom of the Contexts: Active Ensemble Learning for Contextual Anomaly Detection
This addresses the challenge of detecting complex contextual anomalies in datasets with many attributes, which is important for real-world anomaly detection systems, though it appears to be an incremental improvement over existing ensemble and active learning methods.
The paper tackles the problem of identifying the right context for detecting contextual anomalies when true contextual attributes are unknown, proposing WisCon which automatically creates and ensembles multiple contexts with varying importance scores. Experiments show WisCon significantly outperforms existing baselines across seven datasets, supporting the hypothesis that no single perfect context exists for uncovering all contextual anomalies.
In contextual anomaly detection, an object is only considered anomalous within a specific context. Most existing methods for CAD use a single context based on a set of user-specified contextual features. However, identifying the right context can be very challenging in practice, especially in datasets, with a large number of attributes. Furthermore, in real-world systems, there might be multiple anomalies that occur in different contexts and, therefore, require a combination of several "useful" contexts to unveil them. In this work, we leverage active learning and ensembles to effectively detect complex contextual anomalies in situations where the true contextual and behavioral attributes are unknown. We propose a novel approach, called WisCon (Wisdom of the Contexts), that automatically creates contexts from the feature set. Our method constructs an ensemble of multiple contexts, with varying importance scores, based on the assumption that not all useful contexts are equally so. Experiments show that WisCon significantly outperforms existing baselines in different categories (i.e., active classifiers, unsupervised contextual and non-contextual anomaly detectors, and supervised classifiers) on seven datasets. Furthermore, the results support our initial hypothesis that there is no single perfect context that successfully uncovers all kinds of contextual anomalies, and leveraging the "wisdom" of multiple contexts is necessary.