论文部分内容阅读
分类器集成作为机器学习算法设计的一种重要策略,具有高泛化性能的优点,已广泛应用于遥感、生物信息、图像处理等领域,逐渐成为研究热点.以三种常用的集成方法bagging、boosting和stacking为线索,分别从算法思想、算法评价、算法改进与变种、算法应用等角度进行探讨,为构建高准确率分类器集成系统建立指导.最后,对分类器集成算法的未来发展方向进行了展望.
As an important strategy of machine learning algorithm design, classifier integration has the advantages of high generalization performance and has been widely used in remote sensing, biological information, image processing and other fields, and has gradually become a research hotspot.Three common methods of bagging, boosting and stacking as clues, and then discussing from the aspects of algorithm idea, algorithm evaluation, algorithm improvement and variant, algorithm application and so on, and finally establish guidance for constructing high accuracy classifier integrated system.Finally, the future development of classifier integration algorithm Looking forward.