支撑向量机的若干医学应用研究

THE NUMBER OF MEDICAL SUPPORT VECTOR MACHINE APPLIED RESEARCH

在医学领域中常用支撑向量机算法对不同的病情症状进行正确的分类和识别.针对所处理数据的不同特征而采用恰当的算法,选择不同的核函数,能够在很大程度上减少计算量,提高分类和识别的速度.如舌色、苔色识别采用线性和非线性结合的交叉训练法,识别率可达93.87%;肿瘤形状特征分类和肾结石分类均采用非线性算法,所用核函数为高斯核函数,准确率可达95%以上.

Commonly used in the medical field support vector machines for different symptoms of the disease and identify the correct classification.However,the data processing for different characteristics,the algorithm should be used properly,choose a different kernel function,can significantly reduce the amount of computation,thereby increasing speed.Such as the tongue color,fur color identification using a combination of linear and nonlinear cross-training method,recognition rate is 93.87%;tumor classification and shape of the stones are non-linear classification algorithm,the kernel function used for the Gaussian kernel,accuracy rate of up to 95%.