سال انتشار: ۱۳۸۵
محل انتشار: هشتمین همایش ملی صنایع دریایی ایران
تعداد صفحات: ۱۲
K.Azaryoun – Master Degree Student of Iran University of Science and Technology and IT expert of Tidewater Company
M.Analoui – Professor of Iran University of Science and Technology, Computer Department.
K.Jalali – IT Manager of Tidewater Company, Professor of Islamic Azad University
Classifiers are abstract entities which try to label unknown objects from a specified dataset. Each classifier works based on its own accuracy, so sometimes it is preferable to use a collection of classifiers (which is called ensemble) instead of single ones and aggregate the answers of each classifier to obtain the final result. Majority Vote is the most famous approach in this case. In this article we have extended combining classifiers and introduced a new genetic approach for the aggregation phase. As a matter of fact in this article a methodology based on the concept of genetic algorithms (GA) is developed for combining classifiers systems to solve yard allocation problem, on the other hand, we have proposed a genetic based approach for fusion classification to be used as a core of yard management system. Both chief attitudes of genetic algorithms, exploration and exploitation, have been used in order to find the most appropriate label for each dataset member upon the probability for each classifier in the ensemble to give the correct class label of each class. The superiority of the proposed system against some customary approaches has been presented for some real data of Shahid Rajaee port complex.