Poland

Home

Printable Version

Back to Previous Page

Iterative Error-Correction Feedback Perceptron


This command creates the Quantitative (Discriminant Function) Model using current Active Sample and Parameter Sets. The Model is built using TMAIN module from ADAPT. TMAIN develops a linear binary discriminant function using the error-correction feedback principle. It implements what has been termed a linear learning machine, or perceptron, or threshold logic unit. The larger the Dean Zone parameter is set, the more the robust model is created. However, setting the Dead Zone too high might render the model creation impossible. It is important to notice that no Model will be created for a dataset that cannot be 100% correctly classified using a linear equation. Use other discriminant function models in such a case, or improve the dataset by adding more descriptors, or by removing outlier Samples.