Section: New Results

Learning Maximum excluding Ellipsoids from Imbalanced Data with Theoretical Guarantees

Participants: G. Metzler, X. Badiche, B. Belkasmi, E. Fromont, A. Habrard, M. Sebban

[8] addresses the problem of learning from imbalanced data. The authors consider the scenario where the number of negative examples is much larger than the number of positive ones. This work proposes a theoretically-founded method, which learns a set of local ellipsoids centered at the minority class examples while excluding the negative examples of the majority class. This task is addressed from a Mahalanobis-like metric learning point of view, which allows deriving generalization guarantees on the learned metric using the uniform stability framework. The experimental evaluation on classic benchmarks and on a proprietary dataset in bank fraud detection shows the effectiveness of the approach, particularly when the imbalance is huge.