基于深度学习与中医数据的多标签分类算法  

Multi-label Classification Algorithm Based on Deep Learning and TCM Data

作　　者：安伟涛 薛安荣[1] 张宇[1] AN Wei-tao;XUE An-rong;ZHANG Yu(School of Computer Science and Communication Engineering,Jiangsu University,Zhenjiang 212013,China)

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212013

出　　处：《软件导刊》2019年第3期25-29,33,共6页Software Guide

摘　　要：针对中医八纲辨证诊断模型因人工设定参数不准确导致模型训练时间过长、无法收敛和易忽略症状与证型间的一对多关系导致诊断结果存在证型遗漏的问题,提出利用深度置信网络RBM机制,通过对输入的样本特征向量逐层进行拟合获得模型最佳权重与阈值,从而解决参数设定问题;同时采用二元关联多标签分类算法解决一对多关系,提高诊断结果准确率。实验表明,改进后的算法有效可行。In view of the problems that the model training time is too long and can not converge to the model of eight class syndrome differentiation diagnosis model of traditional Chinese medicine and the problem that syndrome omission caused by one-to-many relations between symptoms and syndrome types are ignored,the best weight and threshold of the model are obtained by using the RBM mechanism of the deep confidence network to fit the input initial case eigenvector to get the optimum weight and threshold of the model,thus the parameter setting is solved,and the bi-element association multi-label classification algorithm is adopted to solve the one to many relations.The accuracy of the diagnosis results is improved.Experimental results on syndrome differentiation data of eight TCM syndromes also show the effectiveness of the proposed algorithm.

关 键 词：中医诊断 中医八纲辨证 深度学习 多标签学习 TensorFlow 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]