Agnostic Learning with Ensembles of Classifiers
Joerg D. Wichard, Inst. Molecular Pharmacology, Berlin, Germany.
We present a method for building ensembles of models in order to generate
proper classifiers. The main advantage of our method is an automated model
selection procedure and an automated model parameter estimation. The method
is an extension of the classical K-fold-cross-validation approach. We apply
our method to the five data sets provided at the Agnostic Learning vs. Prior
Knowledge Challenge at the IJCNN 2007.