High-dimensional discriminant analysis. (English) Zbl 1128.62072

Summary: We propose a new discriminant analysis method for high-dimensional data, called High-Dimensional Discriminant Analysis (HDDA). Our approach is based on the assumption that high-dimensional data live in different subspaces with low dimensionality. We therefore propose a new parameterization of the Gaussian model which combines the ideas of dimension reduction and constraints on the model. This parameterization takes into account the specific subspace and the intrinsic dimension of each class to limit the number of parameters to estimate. In addition, it is possible to make additional assumptions on the model to further limit the number of parameters. Our experiments on artificial and real datasets highlight that HDDA is more efficient than classical methods in high-dimensional spaces and small learning datasets.


62H30 Classification and discrimination; cluster analysis (statistical aspects)
65C60 Computational problems in statistics (MSC2010)
62H25 Factor analysis and principal components; correspondence analysis
Full Text: DOI Link


[1] Bellman R., Dynamic Programming (1957)
[2] DOI: 10.2307/2291604 · Zbl 0885.62068
[3] Bouveyron , C. , Girard , S. , Schmid , C. ( 2005 ). High dimensional discriminant analysis . Int. Conf. Appl. Stochastic Mod. Data Anal. Brest , France , pp. 526 – 534 .
[4] Cattell R., Multivariate Behav. Res. 1 pp 140– (1966)
[5] DOI: 10.2307/2288721
[6] DOI: 10.1137/0907013 · Zbl 0614.65043
[7] DOI: 10.2307/2965724 · Zbl 0888.62063
[8] DOI: 10.2307/2289860
[9] DOI: 10.1162/153244303322753616 · Zbl 1102.68556
[10] Hastie T., The Elements of Statistical Learning (2001) · Zbl 0973.62007
[11] Jolliffe I., Principal Component Analysis (1986) · Zbl 1011.62064
[12] DOI: 10.2307/2986198 · Zbl 0821.62032
[13] DOI: 10.1002/0471725293
[14] DOI: 10.1016/S0167-9473(96)00043-6 · Zbl 0875.62266
[15] DOI: 10.1214/aos/1176344136 · Zbl 0379.62005
[16] Scott D., Fifteenth Symp. Interface. pp 173– (1983)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.