一种新的基于深度学习的聚类分析算法  

A New Clustering Analysis Algorithm Based on Deep Learning

作　　者：侯远韶 HOU Yuanshao(Department of Information Engineering,Henan Industry and Trade Vocational College,Zhengzhou 451191,China)

机构地区：[1]河南工业贸易职业学院信息工程系,河南郑州451191

出　　处：《新乡学院学报》2018年第12期21-24,共4页Journal of Xinxiang University

基　　金：河南省科技攻关计划项目(0721002210032)

摘　　要：传统图像分类算法需要大量的数据进行有监督训练,造成了数据冗余进而带来维数灾难,为此提出一种基于深度学习的聚类分析算法。聚类分析利用数据内部簇结构和模式进行分类,不需要对样本进行训练得到先验知识,降低了计算复杂度。引入深度学习对数据内部结构和模式进行特征学习,得到数据的初步聚类,再对初步聚类进行不断优化得到最终的分类效果。实验结果表明,算法很好地解决了信息全面与维数灾难的矛盾,具有良好的实用性和主观一致性。In the traditional image classification algorithm,a large amount of data was required for supervised training,which caused redundancy of data and brought about the problem of dimensionality disaster.Thus,a clustering analysis algorithm based on deep learning was proposed.Clustering analysis was an unsupervised process.It used the internal cluster structure and pattern of data to classify.It didn't need to train the sample to obtain prior knowledge,which reduced the computational complexity.Deep learning was also introduced to study the internal structure and pattern of the data,and the preliminary clustering of the data was obtained.Then the preliminary clustering was continuously optimized to obtain the final classification effect.Experiments showed that the new algorithm solved the contradiction between comprehensive information and dimensionality disaster,and possessed good practicability and subjective consistency.

关 键 词：深度学习 特征提取 聚类分析 无监督学习 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]