Classification learning system based on multi-objective GA and megathermal weather forecat

A new classification learning system based on multi-objective GA is proposed in this paper. Firstly, the continuous attributes of samples are made discretion with a supervised segmentation method, so generalization and intelligibility of machine learning are improved. Moreover, comparison and selection mechanism based on partial order in set theory are infused into multi-objective GA. They enhance the ability to choose better chromosomes. The new algorithm is used to forecast megathermal weather in northern Zhejiang province. The experiment result indicates that it has unique intelligence, higher accuracy.