| Attributs | Valeurs |
|---|
| type
| |
| Is Part Of
| |
| Subject
| |
| Title
| - Neural Network Studies. 3. Variable Selection in the Cascade-Correlation LearningArchitecture
|
| has manifestation of work
| |
| related by
| |
| Author
| |
| Abstract
| - Pruning methods for feed-forward artificial neural networks trained by the cascade-correlation learningalgorithm are proposed. The cascade-correlation algorithm starts with a small network and dynamicallyadds new nodes until the analyzed problem has been solved. This feature of the algorithm removes therequirement to predefine the architecture of the neural network prior to network training. The developedpruning methods are used to estimate the importance of large sets of initial variables for quantitative structure−activity relationship studies and simulated data sets. The calculated results are compared with the performanceof fixed-size back-propagation neural networks and multiple regression analysis and are carefully validatedusing different training/test set protocols, such as leave-one-out and full cross-validation procedures. Theresults suggest that the pruning methods can be successfully used to optimize the set of variables for thecascade-correlation learning algorithm neural networks. The use of variables selected by the elaboratedmethods provides an improvement of neural network prediction ability compared to that calculated usingthe unpruned sets of variables.
|
| article type
| |
| is part of this journal
| |
