A study of the associative pattern classifier method for multi-class processes

Pattern-recognition tasks in machine vision provide solutions for industrial automation and manufacturing processes. These applications are done by extracting data from images and comparing them with well-known data stored, returning a result that helps decide whether the measurement is within a known tolerance. Pattern recognition is an artificial intelligence discipline which is focused to associate a set of features that describes an object with a class or category. Into this field, the associative memories that can be seen as a special class of neural network are used to retrieve altered binary patterns. However, in 2003 was designed the Associative Pattern Classifier (APC), which is an associative memory that is capable to extend this approach to pattern classification field. Several proposals have arisen from APC algorithm; nevertheless and in consequence of its variants, this algorithm is limited to bi-class processes. Moreover, the algorithm has a serious problem when it is configured as a hyper plane classification. The present work solves these drawbacks and it extends the algorithm to multi-class problems. An example of this application is made by using a data base provided from real measurement in the health field.