Skip to Main Content
The Associative Pattern Classifier (CAP) is a novel approach to solve the pattern classification problem. Recent experiments of the behavior of this classifier in different applications have given encouraging results. Due a this evidence, It has been thinking about the existence of a minimum number for which a higher value of samples used in the learning phase of this classifier brings a very low effect over their classification performance. This paper present an empiric way to obtain this minimum number based in the structure of the used database. this method allows us to define a minimum size for the set used in the learning phase of CAP for which the final classification performance will be reasonably stable, optimizing time and computational resources in the process.