Discriminant Analysis for Multiway Data

Gisela Lechuga (1) , Laurent Le Brusquet (1) , Vincent Perlbarg (2) , Louis Puybasset (3) , Damien Galanaud (4) , Arthur Tenenhaus (1)

1 L2S - Laboratoire des signaux et systèmes
2 LIF - Laboratoire d'Imagerie Fonctionnelle
3 Johns Hopkins University School of Medicine [Baltimore]
4 Service de Neuroradiologie [CHU Pitié-Salpêtrière]

A multiway Fisher Discriminant Analysis (MFDA) formulation is presented in this paper. The core of MFDA relies on the structural constraint imposed to the discriminant vectors in order to account for the multiway structure of the data. This results in a more parsimonious model than that of Fisher Discriminant Analysis (FDA) performed on the unfolded data table. Moreover, computational and overfitting issues that occur with high dimensional data are better controlled. MFDA is applied to predict the long term recovery of patients after traumatic brain injury from multi-modal brain Magnetic Resonance Imaging. As compared to FDA, MFDA clearly tracks down the discrimination areas within the white matter region of the brain and provides a ranking of the contribution of the neuroimaging modalities. Based on cross validation, the accuracy of MFDA is equal to 77% against 75% for FDA.

Domains: Engineering Sciences [physics]; Signal and Image processing

Gisela Lechuga, Laurent Le Brusquet, Vincent Perlbarg, Louis Puybasset, Damien Galanaud, et al.. Discriminant Analysis for Multiway Data. Springer Proceedings in Mathematics & Statistics, 115-126, 2016, The Multiple Facets of Partial Least Squares and Related Methods. ⟨hal-01235812⟩