Pynnonen, Seppo (1987). Selection of variables in nonlinear discriminant analysis by information criteria. In Proceedings of the Second International Tampere Conference in Statistics, Department of Mathematical Sciences/Statistics (eds. T. Pukkila and S. Puntanen), pp. 627-236. Department of Mathematical Sciences/Statistics, University of Tampere, Finland

Abstract
In this paper variable selection criteria based on criterion functions, like Akaike's (1973) AIC or Schwarz's (1978) BIC, for normal-theory nonlinear discriminant analysis is proposed. The criteria are derived by using Rao's (1970) additional information principle to evaluate the independent information supplied by variables not in the model given the information of the variables in the model. The behavior of the criteria are illusttrated by simulation experiments.