SORR回归算法在胆固醇测定中的应用

借鉴分类问题的算法,推广到回归问题中去,针对用于分类问题的SOR(successive overrelaxation for support vector)支持向量机算法,提出SORR(successive overrelaxation for support vector regression)支持向量回归算法,并应用于医学上三类血浆脂蛋白(VLDL、LDL、HDL)测定样本中胆固醇的含量。数值实验表明:SORR算法有效,与标准的支持向量回归SVR算法相比,保持了相同的回归精度,提高了学习速度,为临床上测定胆固醇含量提供新的有效方法。

Application of successive overrelaxation for support vector regression arithmetic in cholesterol measurements

The measurement of cholesterol levels in blood has important significance to the clinical diagnosis of heart brain blood vessel disease.If the Cholesterol content is too high or low indicates that the cholesterol metabolism possibly has the barrier,therefore clinical analyzes needs to have the more accurate effective cholesterol content determination method.Drawing lessons from the arithmetic SOB (successive overrelaxation for support vector) used in classification problem that extracted by Mangasarian,we extended to the matter of regression,reform the standard arithmetic of support vector regression,propose successive overrelaxation for support vector regression arithmetic,then present a method that can determine serum cholesterol levels from the measurements of spectral content of a blood sam- ple in medical science.The numerical test has improved that the SOBB( successive overrelaxation for support vector regression)arith- metic is effective and it has the same regression precision compared with the standard support vector regression (SVB) arithmetic.The speed of study is also improved.This paper provides a new method for the cholesterol measurements in clinic.