Attacking DBSCAN for Fun and Profit
Demonstrates how adversaries can subvert DBSCAN clustering by injecting bridge points to merge arbitrary clusters, degrading system performance, and proposes machine learning-based remediation using outlier detection.