[1]钟桂凤,庞雄文,孙道宗.基于差分进化的卷积神经网络的文本分类研究[J].南京师大学报(自然科学版),2022,(01):136-141.[doi:10.3969/j.issn.1001-4616.2022.01.019]
 Zhong Guifeng,Pang Xiongwen,Sun Daozong.Research on Text Classification Based on Convolutional Neural Network of Differential Evolution[J].Journal of Nanjing Normal University(Natural Science Edition),2022,(01):136-141.[doi:10.3969/j.issn.1001-4616.2022.01.019]
点击复制

基于差分进化的卷积神经网络的文本分类研究()
分享到:

《南京师大学报(自然科学版)》[ISSN:1001-4616/CN:32-1239/N]

卷:
期数:
2022年01期
页码:
136-141
栏目:
·计算机科学与技术·
出版日期:
2022-03-15

文章信息/Info

Title:
Research on Text Classification Based on Convolutional Neural Network of Differential Evolution
文章编号:
1001-4616(2022)01-0136-06
作者:
钟桂凤1庞雄文2孙道宗3
(1.广州理工学院计算机科学与工程学院,广东 广州 510540)(2.华南师范大学计算机学院,广东 广州 530631)(3.华南农业大学 电子工程学院,广东 广州 510642)
Author(s):
Zhong Guifeng1Pang Xiongwen2Sun Daozong3
(1.College of Computer Science & Engineering,Guangzhou Institute of Science and Technology,Guangzhou 530631,China)(2.School of Computer Science,South China Normal University,Guangzhou 530631,China)(3.School of Electronic Engineering,South China Agricultural University,Guangzhou 530631,China)
关键词:
分类差分进化卷积神经网络缩放因子卷积核
Keywords:
classificationdifferential evolutionconvolution neural networkscaling factorconvolution kernel
分类号:
TP391.1
DOI:
10.3969/j.issn.1001-4616.2022.01.019
文献标志码:
A
摘要:
为了提高文本分类的性能,采用差分进化的卷积神经网络(convolutional neural network,CNN)算法进行分类. 首先随机设置CNN结构参数,然后采用差分进化算法优化参数,通过交叉和选择等操作选择不断进化获得最优个体,为增强差分优化的适用性,将缩放因子变化与进化代数相关联,解决了因为缩放因子设置不合理而造成优化等级不高的问题. 卷积神经网络采用经过差分优化后的权重和阈值对文本进行分类训练,以获得稳定的文本分类结果. 实验证明,通过合理设置差分进化交叉速率和卷积神经网络的卷积核尺寸,能够获得较好的分类准确率性能,RMSE值更低,在文本分类中的适用度高.
Abstract:
In order to improve the performance of text classification,differential evolution convolutional neural network(CNN)algorithm was used for text classification. Firstly,the structural parameters of CNN were set randomly,then the parameters were optimized by differential evolution algorithm,and the optimal individuals were obtained by continuous evolution through crossover and selection. In order to enhance the applicability of differential optimization,the change of scaling factor was associated with evolutionary algebra,which solved the problem of low optimization level caused by unreasonable setting of scaling factor. Convolutional neural network used the weight and threshold after differential optimization to train text classification,so as to obtain stable text classification results. Experimental results showed that by setting the crossover rate of differential evolution and convolution kernel size of convolution neural network reasonably,better classification accuracy performance can be obtained,RMSE value was lower,and applicability in text classification was high.

参考文献/References:

[1] TANG X C,DAI Y S,XIANG Y P,et al. Feature selection based on feature interactions with application to text categorization[J]. Expert systems with applications,2019,120:207-216.
[2]于游,付钰,吴晓平. 中文文本分类方法综述[J]. 网络与信息安全学报,2019(5):1-8.
[3]郭超磊,陈军华. 基于SA-SVM的中文文本分类研究[J]. 计算机应用与软件,2019(3):277-281.
[4]SHU J B,SHEN X X,LIU H,et al. A content-based recommendation algorithm for learning resources[J]. Multimedia systems,2018,24(2):163-173.
[5]孙越泓,王丹. 基于新约束集成的差分进化算法[J]. 南京师大学报(自然科学版),2019,42(4):1-11.
[6]王永安,赵阳,蓝雨晨,等. 基于进化差分算法的环形电感建模及应用[J]. 南京师范大学学报(工程技术版),2020,20(3):32-37.
[7]SLW A,FM A,TFN B,et al. Insights into the effects of control parameters and mutation strategy on self-adaptive ensemble-based differential evolution-ScienceDirect[J]. Information sciences,2020,514:203-233.
[8]QLAB C,SD C,BJVW D,et al. Double-layer-clustering differential evolution multimodal optimization by speciation and self-adaptive strategies-science direct[J]. Information sciences,2021,545:465-486.
[9]AMIT K S,SANDEEP C,DEVESH K S. Sentimental short sentences classification by using cnn deep learning model with fine tuned Word2Vec[J]. Procedia computer science,2020,167:1139-1147.
[10]PAN R L,YU C,ZHAO W,et al. Multi-channel Sliced Deep RCNN with residual network for text classification[J]. Chinese journal of electronics,2020,29(5):92-98.
[11]LIU Y,SUEN C Y,LIU Y,et al. Scene classification using Hierarchical wasserstein CNN[J]. IEEE Transactions on geoscience and remote sensing,2019,57(5):2494-2509.
[12]沈浩,江臣,陈宇文,等. 基于深度学习的钢桁架桥螺栓病害智能识别方法[J]. 南京工业大学学报(自然科学版),2020,42(5):608-614.
[13]NADERALVOJOUD B,SEZER E A. Term uation metrics in imbalanced text categorization[J]. Natural Language engineering,2019,26(1):1-17.
[14]梁柯,李健,陈颖雪,等. 基于朴素贝叶斯的文本情感分类及实现[J]. 智能计算机与应用,2019(5):150-153,157.
[15]钱鹏,陆金桂. 基于PSO-BP神经网络的红外无损检测缺陷定量识别[J]. 南京工业大学学报(自然科学版),2019,41(4):501-507.
[16]田园,原野,刘海斌,等. 基于BERT预训练语言模型的电网设备缺陷文本分类[J]. 南京理工大学学报,2020,44(4):446-453.

相似文献/References:

[1]刘钦普.国内低碳城市的概念及评价指标体系研究评述[J].南京师大学报(自然科学版),2014,37(02):1.
 Liu Qinpu.Review of Researches on Evaluation Index Systems of LowCarbon City in China[J].Journal of Nanjing Normal University(Natural Science Edition),2014,37(01):1.
[2]贾 涛,韩 萌,王少峰,等.数据流决策树分类方法综述[J].南京师大学报(自然科学版),2019,42(04):49.[doi:10.3969/j.issn.1001-4616.2019.04.008]
 Jia Tao,Han Meng,Wang Shaofeng,et al.Survey of Decision Tree Classification Methods over Data Streams[J].Journal of Nanjing Normal University(Natural Science Edition),2019,42(01):49.[doi:10.3969/j.issn.1001-4616.2019.04.008]

备注/Memo

备注/Memo:
收稿日期:2021-05-07.
基金项目:国家自然科学基金项目(31101077)、2020年度广东省高校科研项目(2020GXJK201).
通讯作者:钟桂凤,讲师,研究方向:数据分析与挖掘,人工智能,机器学习. E-mail:109488818@qq.com
更新日期/Last Update: 1900-01-01