We propose a semiparametric linear programming discriminant (SLPD) rule for high dimensional discriminant analysis under a semiparametric model. As an extension, we further propose a two-stage SLPD (TSLPD) rule, which can have better classification performance under mild sparsity assumptions.
- Bayes rule
- Linear discrimination analysis
- Monotone transformation
- Semiparametric discriminant analysis
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty