摘要
胎重的准确预测在胎儿监护中具有重要作用,为提高胎重预测的准确性,提出一种基于模糊支持向量机的预测方法。通过在支持向量机中引入模糊逻辑,抑制由测量误差造成的异常数据对预测模型训练的影响,提高了胎重预测对参数测量误差的鲁棒性。对600例数据构成的训练集和150例数据构成的测试集进行应用,比较了模糊支持向量机和以前的回归方法、误差反向传递神经网络、支持向量机在胎重预测中的性能。结果表明:与其它方法相比,模糊支持向量机能获得更准确的胎重估计。
Accurate fetal weight estimation has paramount effect in prenatal care. To improve the accuracy of fetal weight estimation, a novel method is proposed based on fuzzy support vector machine. Fuzzy logic is introduced into the support vector regression to suppress the effect of outliers caused by measurement errors on the model training, which leads to the improvement of the estimation robustness to random measurement errors. A training set with 600 fetuses and a validating set with 150 fetuses are utilized to compare the performances of regression formulas, backpropagation network, support vector regression, and fuzzy support vector regression in fetal weight estimation. Experimental results indicate that fuzzy support vector regression can obtain more accurate fetal weight estimation than other methods.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2008年第11期2241-2246,共6页
Chinese Journal of Scientific Instrument
基金
国家重点基础研究规划基金(2006CB705707)
国家自然科学基金(30570488)
上海市重点学科建设项目(B112)
复旦大学研究生创新基金(EYH1220001)资助项目
关键词
胎重估计
回归方法
误差反向传递神经网络
支持向量机
模糊支持向量机
fetal weight estimation
regression method
back-propagation network
support vector regression
fuzzy support vector regression
