Stacking for Ensembles of Local Experts in Metabonomic Applications

  • Kai Lienemann
  • Thomas Plötz
  • Gernot A. Fink
Conference paper

DOI: 10.1007/978-3-642-02326-2_50

Volume 5519 of the book series Lecture Notes in Computer Science (LNCS)
Cite this paper as:
Lienemann K., Plötz T., Fink G.A. (2009) Stacking for Ensembles of Local Experts in Metabonomic Applications. In: Benediktsson J.A., Kittler J., Roli F. (eds) Multiple Classifier Systems. MCS 2009. Lecture Notes in Computer Science, vol 5519. Springer, Berlin, Heidelberg

Abstract

Recently, Ensembles of local experts have successfully been applied for the automatic detection of drug-induced organ toxicities based on spectroscopic data. For suitable Ensemble composition an expert selection optimization procedure is required that identifies the most relevant classifiers to be integrated. However, it has been observed that Ensemble optimization tends to overfit on the training data. To tackle this problem we propose to integrate a stacked classifier optimized via cross-validation that is based on the outputs of local experts. In order to achieve probabilistic outputs of Support Vector Machines used as local experts we apply a sigmoidal fitting approach. The results of an experimental evaluation on a challenging data set from safety pharmacology demonstrate the improved generalizability of the proposed approach.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Kai Lienemann
    • 1
  • Thomas Plötz
    • 1
  • Gernot A. Fink
    • 1
  1. 1.Intelligent Systems GroupTU Dortmund UniversityGermany