FEMDA: A Unified Framework for Discriminant Analysis

Although linear and quadratic discriminant analysisQuadratic discriminant analysis (QDA) are widely recognized classical methods, they can encounter significant challenges when dealing with non-Gaussian distributions or contaminated datasets. This is primarily due to their reliance on the Gaussian assumption, which lacks robustness. We first explain and review the classical methods to address this limitation and then present a novel approach that overcomes these issues. In this new approach, the model considered is an arbitrary Elliptically Symmetrical (ES) distribution per cluster with its own arbitrary scaleScale parameter. This flexible model allows for potentially diverse and independent samples that may not follow identical distributions. By deriving a new decision rule, we demonstrate that maximum-likelihood parameter estimationParameter estimation and classificationClassification are simple, efficient, and robust compared to state-of-the-art methods.