An Adjusted Nearest Neighbor Algorithm Maximizing the F-Measure from Imbalanced Data

Rémi Viola (1, 2, 3) , Rémi Emonet (1) , Amaury Habrard (1) , Guillaume Metzler (1) , Sébastien Riou (2, 3) , Marc Sebban (1)

1 LHC - Laboratoire Hubert Curien
2 DGFiP - Direction Générale des Finances Publiques
3 Ministère de l'économie et des finances

In this paper, we address the challenging problem of learning from imbalanced data using a Nearest-Neighbor (NN) algorithm. In this setting, the minority examples typically belong to the class of interest requiring the optimization of specific criteria, like the F-Measure. Based on simple geometrical ideas, we introduce an algorithm that reweights the distance between a query sample and any positive training example. This leads to a modification of the Voronoi regions and thus of the decision boundaries of the NN algorithm. We provide a theoretical justification about the weighting scheme needed to reduce the False Negative rate while controlling the number of False Positives. We perform an extensive experimental study on many public imbalanced datasets, but also on large scale non public data from the French Ministry of Economy and Finance on a tax fraud detection task, showing that our method is very effective and, interestingly, yields the best performance when combined with state of the art sampling methods.

Mots clés: Machine Learning; Nearest Neighbor; F- Measure; Imbalanced Classification

Domaines: Intelligence artificielle [cs.AI]

Rémi Viola, Rémi Emonet, Amaury Habrard, Guillaume Metzler, Sébastien Riou, et al.. An Adjusted Nearest Neighbor Algorithm Maximizing the F-Measure from Imbalanced Data. International Conference on Tools with Artificial Intelligence (ICTAI), Nov 2019, Portland, Oregon, United States. ⟨hal-02318868⟩