Suspicion-Free Adversarial Attacks on Clustering Algorithms

Clustering algorithms are used in a large number of applications and play an important role in modern machine learningyet, adversarial attacks on clustering algorithms seem to be broadly overlooked unlike supervised learning. In this paper, we seek to bridge this gap by proposing a black-box adversarial attack for clustering models for linearly separable clusters. Our attack works by perturbing a single sample close to the decision boundary, which leads to the misclustering of multiple unperturbed samples, named spill-over adversarial samples. We theoretically show the existence of such adversarial samples for the K-Means clustering. Our attack is especially strong as (1) we ensure the perturbed sample is not an outlier, hence not detectable, and (2) the exact metric used for clustering is not known to the attacker. We theoretically justify that the attack can indeed be successful without the knowledge of the true metric. We conclude by providing empirical results on a number of datasets, and clustering algorithms. To the best of our knowledge, this is the first work that generates spill-over adversarial samples without the knowledge of the true metric ensuring that the perturbed sample is not an outlier, and theoretically proves the above. * Equal contribution and the names are in alphabetical order of second names.