【摘要】 本文讨论了多分类器组合中的分类器选择问题,提出一种基于遗传算法的分类器选择算法,此算法可以快速选出有效的分类器参与组合。文中给出了指定分类器数目和任意分类器数目两种情况下分类器选择的算法。最后在CENPARMI手写体数字数据库上验证了我们的算法和结沦。实验结果表明,此种分类器选择算法具有较好的性能。更多还原

【Abstract】 The classifier selection problem in multi-classifier combination is discussed in this paper. The classifier selection algorithms based on GA is proposed, which can finds out the best classifiers to combine. A classifier selection algorithm of fixed number and unfixed number classifier combination is given. At the end of the paper, the experiments based on CEMPARMI handwritten digit database are given to prove our conclusion. The result shows that the performance of the algorithm is good.更多还原