Generalization theory of two-part code MDL estimator
Published on Aug 13, 20083928 Views
I will present a finite-sample generalization analysis of two-part code MDL estimator. This method selects a model that minimizes the sum of the model description length plus the data description leng