Regularized Negative Correlation Learning and its application to Cheminformatics

Huanhuan Chen

School of Computing, University of Leeds

Driven by new regulations and animal welfare, the need to develop in silico models has increased recently as alternative approaches to safety assessment of chemicals without animal testing. This talk describesa novel machine learning ensemble approach to building an in silico model for the prediction of the Ames test mutagenicity, one of a battery of the most commonly used experimental in vitro and in vivo genotoxicity tests for safety evaluation of chemicals. Regularized negative correlation learning (RNCL) was developed based on neural network ensembles and statistical Bayesian inference. RNCL not only considers the accuracy of individual networks in ensemble, but also pays more attention to the cooperation among these networks and the anti-noise ability. Furthermore, we present multi-objective algorithm to train RNCL. The empirical experiment suggests that RNCL is an effective ensemble approach for predicting the Ames test mutagenicity of chemicals.