FEMDA: Une méthode de classification robuste et flexible
This addresses robustness issues in classification for scenarios with heterogeneous and non-identically distributed data, though it appears incremental as an extension of discriminant analysis.
The paper tackles the problem of LDA and QDA suffering from non-Gaussian distributions and contaminated datasets by proposing FEMDA, a new discriminant analysis technique that models each data point with its own arbitrary Elliptically Symmetrical distribution and scale parameter, resulting in a simple, fast, and robust decision rule compared to state-of-the-art methods.
Linear and Quadratic Discriminant Analysis (LDA and QDA) are well-known classical methods but can heavily suffer from non-Gaussian distributions and/or contaminated datasets, mainly because of the underlying Gaussian assumption that is not robust. This paper studies the robustness to scale changes in the data of a new discriminant analysis technique where each data point is drawn by its own arbitrary Elliptically Symmetrical (ES) distribution and its own arbitrary scale parameter. Such a model allows for possibly very heterogeneous, independent but non-identically distributed samples. The new decision rule derived is simple, fast, and robust to scale changes in the data compared to other state-of-the-art method