Repository | Book | Chapter
(2002) Progress in discovery science, Dordrecht, Springer.
Constructing inductive applications by meta-learning with method repositories
Hidenao Abe , Takahira Yamaguchi
pp. 576-585
Here is presented CAMLET that is a platform for automatic composition of inductive applications with method repositories that organize many inductive learning methods. CAMLET starts with constructing a basic design specification for inductive applications with method repositories and data type hierarchy that are specific to inductive learning algorithms. After instantiating the basic design with a given data set into a detailed design specification and then compiling it into codes, CAMLET executes them on computers. CAMLET changes the constructed specification until it goes beyond the goal accuracy given from a user. After having implemented CAMLET on UNIX platforms with Perl and C languages, we have done the case studies of constructing inductive applications for eight different data sets from the StatLog project and have compared the accuracies of the inductive applications composed by CAMLET with all the accuracies from popular inductive learning algorithms. The results have shown us that the inductive applications composed by CAMLET take the first accuracy on the average.
Publication details
Full citation:
Abe, H. , Yamaguchi, T. (2002)., Constructing inductive applications by meta-learning with method repositories, in S. Arikawa & A. Shinohara (eds.), Progress in discovery science, Dordrecht, Springer, pp. 576-585.
This document is unfortunately not available for download at the moment.