为了分析多类支持向量机(Multi-category support vector machines,M-SVMs)的推广性能,对常用的M-SVMs算法加以概述,推导、总结了理论推广误差公式。对于给定的样本集,可以设计合理的编码来提高ECOCSVMs的推广性能,通过构造合理的层次...为了分析多类支持向量机(Multi-category support vector machines,M-SVMs)的推广性能,对常用的M-SVMs算法加以概述,推导、总结了理论推广误差公式。对于给定的样本集,可以设计合理的编码来提高ECOCSVMs的推广性能,通过构造合理的层次结构来提高H-SVMs推广性能,其余M-SVMs算法的推广性能均取决于样本空间。研究结果为有效使用M-SVMs提供了依据,为改进M-SVMs指明了方向。展开更多
To recognize the presence of the headstream of gushing water in coal mines, the SVM (Support Vector Ma- chine) was proposed to analyze the gushing water based on hydrogeochemical methods. First, the SVM model for head...To recognize the presence of the headstream of gushing water in coal mines, the SVM (Support Vector Ma- chine) was proposed to analyze the gushing water based on hydrogeochemical methods. First, the SVM model for head- stream analysis was trained on the water sample of available headstreams, and then we used this to predict the unknown samples, which were validated in practice by comparing the predicted results with the actual results. The experimental results show that the SVM is a feasible method to differentiate between two headstreams and the H-SVMs (Hierachical SVMs) is a preferable way to deal with the problem of multi-headstreams. Compared with other methods, the SVM is based on a strict mathematical theory with a simple structure and good generalization properties. As well, the support vector W in the decision function can describe the weights of the recognition factors of water samples, which is very important for the analysis of headstreams of gushing water in coal mines.展开更多
文摘为了分析多类支持向量机(Multi-category support vector machines,M-SVMs)的推广性能,对常用的M-SVMs算法加以概述,推导、总结了理论推广误差公式。对于给定的样本集,可以设计合理的编码来提高ECOCSVMs的推广性能,通过构造合理的层次结构来提高H-SVMs推广性能,其余M-SVMs算法的推广性能均取决于样本空间。研究结果为有效使用M-SVMs提供了依据,为改进M-SVMs指明了方向。
基金Project 40401038 supported by the National Natural Science Foundation of China and 2003047 by the Top 100 Outstanding Doctoral Dissertation Foun-dation of China
文摘To recognize the presence of the headstream of gushing water in coal mines, the SVM (Support Vector Ma- chine) was proposed to analyze the gushing water based on hydrogeochemical methods. First, the SVM model for head- stream analysis was trained on the water sample of available headstreams, and then we used this to predict the unknown samples, which were validated in practice by comparing the predicted results with the actual results. The experimental results show that the SVM is a feasible method to differentiate between two headstreams and the H-SVMs (Hierachical SVMs) is a preferable way to deal with the problem of multi-headstreams. Compared with other methods, the SVM is based on a strict mathematical theory with a simple structure and good generalization properties. As well, the support vector W in the decision function can describe the weights of the recognition factors of water samples, which is very important for the analysis of headstreams of gushing water in coal mines.
