SORR回归算法在胆固醇测定中的应用  

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

作　　者：刘太安[1] 杨柏翠[1] 李涵[1] 韩威[1] 

机构地区：[1]山东科技大学信息工程系,山东泰安271019

出　　处：《计算机与应用化学》2007年第9期1163-1165,共3页Computers and Applied Chemistry

基　　金：国家自然科学基金(10171055);山东省自然科学基金(2007ZRB019FK)

摘　　要：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。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 SOR （ 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 sample in medical science. The numerical test has improved that the SORR（ successive overrelaxation for support vector regression）arithmetic is effective and it has the same regression precision compared with the standard support vector regression （SVR） arithmetic. The speed of study is also improved. This paper provides a new method for the cholesterol measurements in clinic.

关 键 词：SOR SVR SORR 回归算法 胆固醇测定 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]