A Dynamic Classification Method for the Discrimination of Evolving Data

Classes issued of evolving systems are dynamic and their characteristics vary over the time. Assigning a pattern to a class is achieved using a classifier. Therefore, the classifier parameters must be adapted online in order to take into account the temporal changes in the classes' characteristics. This adaptation is based only on the recent and useful information carried out by the new incoming classified patterns. In this paper, we propose to develop the classification method Fuzzy Pattern Matching (FPM) to be operant in the case of dynamic classes. This development is based on the use of an incremental algorithm to follow the accumulated gradual temporal changes of classes' characteristics after the classification of each new pattern. When these changes reach a suitable predefined threshold, the classifier parameters are adapted online using the recent and representative patterns. Keywords-- Classification, dynamic patterns, evolving systems, pattern recognition.

• Type: Journal
• Year: 2009
• Where: EUSFLAT
• Authors: Laurent Hartert, Moamar Sayed Mouchaweh, Patrice Billaudel
• Keywords: Classifier Parameters, EUSFLAT 2009, Evolving Systems, Fuzzy Logic, Temporal Changes