«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

31

期数:

2021年07期

页码:

7-12

栏目:

人工智能

出版日期:

2021-07-10

文章信息/Info

Title:

Prediction Model of Protein-protein Interactions Based on Semi-supervised Learning

文章编号:

1673-629X(2021)07-0007-06

作者:

安计勇¹; 2; 闫子骥¹; 2

1. 中国矿业大学 矿山数字化教育部工程研究中心,江苏 徐州 221000;

2. 中国矿业大学 计算机科学与技术学院,江苏 徐州 221000

Author(s):

AN Ji-yong¹; 2; YAN Zi-ji¹; 2

1. Engineering Research Center of Mine Digitalization of Ministry of Education,China University of Mining and Technology,Xuzhou 221000,China;

2. School of Computer Science and Technology,China University of Mining and Technology,Xuzhou 221000,China

关键词:

相关向量机; 半监督学习; 自训练; AP 聚类; Renyi 熵; 分类预测

Keywords:

relevance vector machine; semi-supervised learning; self-training; AP clustering; Renyi entropy; classification prediction

分类号:

TP311. 13

DOI:

10. 3969 / j. issn. 1673-629X. 2021. 07. 002

摘要:

基于有监督学习的预测模型在预测过程中存在以下缺陷:一是过分依赖训练集中有标签样本的数量, 导致分类精度受有标签样本数量多少的制约;二是其预测分类一次完成, 导致大量的无标签样本无法用来修正分类器的预测精度,大量数据信息被浪费,从而影响分类性能。 针对以上问题,该文提出一种基于 AP 聚类与 Renyi 熵融合的自训练半监督相关向量机分类预测模型。 该模型通过 AP 聚类分析与 Renyi 熵来共同标记无标签样本的标签类别,筛选置信度高的无标签样本扩充原有训练集进行自训练迭代分类,降低噪声数据对分类器预测精度的影响,构造出了性能最优的基于半监督学习的蛋白质相互作用预测模型。 通过在 M. musculus、H. pylori 和 H. sapiens 蛋白质相互作用数据集上的实验验证,证明了提出的半监督分类预测模型的有效性。

Abstract:

In the prediction process,the disadvantage of the prediction model based on supervised learning are as follows:firstly,due to over-dependence on the number of labeled samples in the training set,the classification accuracy is limited by the number of labeled samples. Secondly,its prediction classification is completed once,resulting in a large number of unlabeled samples that cannot be used to amend the prediction accuracy of the classifier, and a large amount of data information is wasted, thus affecting the classification perform-ance. In view of the above problems,we propose a new self-training semi-supervised classification prediction model of RVM based on AP clustering and Renyi entropy fusion. This model can greatly reduce the influence of noise data on the prediction accuracy of classifier by using AP clustering and Renyi entropy fusion to assign labels for unlabeled samples. The semi-supervised classifier with optimal performance was constructed through adding the unlabeled samples with high degree of confidence to the training set and executing the self-training iteration classification with the expanded training set. It is demonstrated that the proposed prediction model is effective by experimenting validation on M. musculus,H. pylori and H. sapiens datasets.

常用功能

更新日期/Last Update: 2021-07-10