基于t分布随机邻域嵌入的阿尔茨海默症诊断模型

发布时间：2018-06-19 03:11

本文选题：支持向量机 + t分布随机邻域嵌入　；参考：《厦门大学学报(自然科学版)》2017年01期

【摘要】：对大脑皮层厚度数据进行建模从而实现阿尔茨海默症的诊断.在训练样本少,数据复杂且非线性的情况下,相比于BP神经网络和k最近邻等算法,支持向量机算法表现出更优良的特性.针对支持向量机算法受数据高维度的影响,将t分布随机邻域嵌入算法引入到支持向量机模型.t分布随机邻域嵌入算法既能撷取原始高维数据的局部信息,也能揭示全局结构.t分布随机邻域嵌入算法先将这些非线性数据降维到低维空间,支持向量机算法再将这数据映射到新的高维空间,通过寻找最佳分类超平面的方法,使分类效果达到最佳水平.最后将集成学习算法AdaBoost的思想融入模型,可以使模型的分类准确率得到提升,而且变得鲁棒性更强.
[Abstract]:The cortical thickness data are modeled to diagnose Alzheimer's disease. When the training sample is small, the data is complex and nonlinear, the support vector machine (SVM) algorithm has better performance than BP neural network and k-nearest neighbor algorithm. Because the support vector machine (SVM) algorithm is affected by the high dimension of the data, the t-distribution random neighborhood embedding algorithm is introduced to the support vector machine model. The t-distribution random neighborhood embedding algorithm can retrieve the local information of the original high-dimensional data. It can also be revealed that the global structure. T distribution random neighborhood embedding algorithm first reduces the dimension of these nonlinear data to the low dimensional space, then the support vector machine algorithm maps the data to a new high dimensional space, and by looking for the best classification method of hyperplane, the algorithm of support vector machine maps the data to a new high dimensional space. Make the classification effect reach the best level. Finally, the idea of integrated learning algorithm AdaBoost is integrated into the model, which can improve the classification accuracy of the model and become more robust.
【作者单位】：厦门大学信息科学与技术学院;
【分类号】：R749.16;TP18

【相似文献】