Neural Networks in QSAR and Drug Design
Neural Networks in QSAR and Drug Design
Parameter Selection in Particle Swarm Optimization
EP '98 Proceedings of the 7th International Conference on Evolutionary Programming VII
| Hi-index | 0.00 |
The development of quantitative structure-activity relationship (QSAR) models for computer-assisted drug design is an established technique in the pharmaceutical industry. QSAR models provide a framework for predicting a compound's biological activity based on its chemical structure or properties, and can significantly reduce the time to discover a new drug. In this paper we describe the application of a new optimization technique, particle swarms, to develop QSAR models based on k-nearest neighbor and kernel regression. Particle swarms is a population-based stochastic method that has been used successfully for feature selection. Each individual explores the feature space guided by its previous success and that of its neighbors. Success is measured by the predictivity of the resulting model as determined by k-nearest neighbor and kernel regression. The swarm flies through the feature space in search of the global minimum, guided by the regression error. The technique is evaluated using well-known QSAR data sets and compared to other machine learning techniques.