摘要
根据多模型可以改善模型估计精度,提高泛化性的思想,提出一种基于改进加权粗糙集的多模型软测量建模方法。加权粗糙集可以有效地处理不平衡数据的分类问题,但是传统的样本权重选择方法缺乏整体考虑,容易引起分类器整体精度的下降。通过向加权粗糙集引入类别权重,得到了一种基于最小风险贝叶斯决策理论的加权粗糙集决策算法,并利用AdaBoostM2算法寻优样本权重及类别权重。通过上述方法构建的最小风险加权粗糙集分类器,有效地提高了分类精度,从而保证了各个子模型的可靠性。
According to the idea that multi-models could improve the estimated accuracy and generalization,a soft-sensing method with multiple models based on an improved weighted rough set was presented.Weighted rough set was effective for class imbalance learning,but it might result in the drop of the classification precision due to lacking fully consideration of selecting sample weighting function.By introducing label weighting function into weighted rough set,a Bayes decision algorithm based on minimum risk was presented.Meanwhile,AdaBoostM2 algorithm was used to search optimization of sample weighting function and label weighting function.The minimum risk weighted rough set classifier which constructed by optimum parameters,effectively boosts the classification accuracy of classifier and ensures the reliability of sub model.
出处
《化工自动化及仪表》
CAS
北大核心
2010年第1期11-15,共5页
Control and Instruments in Chemical Industry
基金
国家自然科学基金资助项目(60674092)
江苏省高技术研究项目(BG20060010)
江南大学创新团队发展计划资助项目
