Fair Hierarchical Clustering

Sara Ahmadian, Alessandro Epasto, Marina Knittel, Ravi Kumar, Mohammad Mahdian, Benjamin Moseley, Philip Pham, Sergei Vassilvitskii, Yuyan Wang

Abstract:

As machine learning has become more prevalent, researchers have begun to recog-nize the necessity of ensuring machine learning systems are fair. Recently, there hasbeen an interest in defining a notion of fairness that mitigates over-representationin traditional clustering.

In this paper we extend this notion to hierarchical clustering, where the goal is torecursively partition the data to optimize a specific objective. For various naturalobjectives, we obtain simple, efficient algorithms to find a provably good fairhierarchical clustering. Empirically, we show that our algorithms can find a fairhierarchical clustering, with only a negligible loss in the objective.