[1]张永梅,付昊天,张睿.采用BP算法和深度SAE网络的学生综合能力评价方法[J].华侨大学学报(自然科学版),2018,39(5):774-780.[doi:10.11830/ISSN.1000-5013.201707010]
　ZHANG Yongmei,FU Haotian,ZHANG Rui.Evaluation Method for Students’ Comprehensive Abilities Using Deep SAE Networks and BP Algorithm[J].Journal of Huaqiao University(Natural Science),2018,39(5):774-780.[doi:10.11830/ISSN.1000-5013.201707010]
点击复制

采用BP算法和深度SAE网络的学生综合能力评价方法()
分享到：

《华侨大学学报（自然科学版）》[ISSN:1000-5013/CN:35-1079/N]

卷:
第39卷
期数:
2018年第5期
页码:
774-780
栏目:
出版日期:
2018-09-20

文章信息/Info

Title:
Evaluation Method for Students’ Comprehensive Abilities Using Deep SAE Networks and BP Algorithm
文章编号:
1000-5013(2018)05-0774-07
作者:
张永梅1;  付昊天1;  张睿2
1. 北方工业大学 计算机学院, 北京 100144;2. 太原科技大学 计算机科学与技术学院, 山西 太原 030024
Author(s):
ZHANG Yongmei1;  FU Haotian1;  ZHANG Rui2
1. College of Computer Science and Technology, North China University of Technology, Beijing 100144, China; 2. School of Computer Science and Technology, Taiyuan University of Science and Technology, Taiyuan 030024, China
关键词:
反向传播算法;  深度神经网络;  堆栈式自编码器;  综合能力评价
Keywords:
back propagation algorithm;  deep neural network;  stacked autoencoder;  comprehensive ability evaluation
分类号:
TP183
DOI:
10.11830/ISSN.1000-5013.201707010
文献标志码:
A
摘要:
针对现有评价方法需人工提取特征且评价准确率低的问题,提出基于反向传播(BP)算法的深度堆栈编码器(SAE)网络的学生综合能力评价方法.通过SAE网络对输入的学生各项指标成绩进行无监督训练,将SAE学习到的特征结合相应的样本标签,利用柔性最大值分类器(Softmax)进行有监督式分类.采用BP算法进行反向传播调整隐层权重,优化整个模型,以避免过拟合现象的发生.结果表明:该评价方法有利于解决需对传统神经网络进行人工提取和分析特征的问题,可提高评价结果的准确率.
Abstract:
Since the existing evaluation methods need to extract features manually, and the evaluation accuracy is lower. The paper proposes an evaluation method for students’ comprehensive abilities based on deep SAE Networks and back propagation(BP)algorithm. The method adopts stacked autoencoder(SAE)network for students’ various scores with unsupervised training, and utilizes the features extracted by SAE network and the corresponding sample labels to train the network via supervised Softmax classifier. The method uses BP algorithm to adjust the weights for hidden layers and optimize the entire model, and avoids the occurrence of over-fitting. The method can help to solve the problem of manual extraction and analysis features for traditional neural networks, and improve the evaluation accuracy.

参考文献/References:

[1] SAMUEL O W,ASOGBON G M,SANGAIAH A K,et al.An integrated decision support system based on ANN and Fuzzy_AHP for heart failure risk prediction[J].Expert Systems with Applications,2017,68(2017):163-172.DOI:10.1016/j.eswa.2016.10.020.
[2] DENG Zhenyun,ZHU Xiaoshu,CHENG Debo,et al.Efficient kNN classification algorithm for big data[J].Neurocomputing,2016,195(C):143-148.DOI:10.1016/j.neucom.2015.08.112.
[3] 杨玮,岳婷,曹薇,等.采用BP神经网络的猪肉冷链物流预警模型与仿真[J].华侨大学学报(自然科学版),2015,36(5):511-516.DOI:10.11830/ISSN.1000-5013.2015.05.0511.
[4] FARIAS G,DORMIDO-CANTO S,VEGA J,et al.Automatic feature extraction in large fusion databases by using deep learning approach[J].Fusion Engineering and Design,2016,112:979-983.DOI:10.1016/j.fusengdes.2016.06.016.
[5] WANG Yasi,YAO Hongxun,ZHAO Sicheng.Auto-encoder based dimensionality reduction[J].Neurocomputing,2016,184(C):232-242.DOI:10.1016/j.neucom.2015.08.104.
[6] JIANG Wenhao,GAO Hongchang,CHUNG Fulai,et al.The L2,1-norm stacked robust autoencoders for domain adaptation[C]//Thirtieth AAAI Conference on Artificial Intelligence.Phoenix:AAAI,2016:1723-1729.
[7] CAO Lele,HUANG Wenbing,SUN Fuchun.Building feature space of extreme learning machine with sparse denoising stacked-autoencoder[J].Neurocomputing,2016,174(Part A):60-71.DOI:10.1016/j.neucom.2015.02.096.
[8] 尚敬文,王朝坤,辛欣,等.基于深度稀疏自动编码器的社区发现算法[J].软件学报,2017,28(3):648-662.DOI:10.13328/j.cnki.jos.005165.
[9] 李阳辉,谢明,易阳.基于降噪自动编码器及其改进模型的微博情感分析[J].计算机应用研究,2017,34(2):373-377.DOI:10.3969/j.issn.1001-3695.2017.02.012.
[10] OUYANG Yuanxin,LIU Wenqi,RONG Wenge,et al.Autoencoder-based collaborative filtering[C]//International Conference on Neural Information Processing.Kuching:Springer,2014:284-291.DOI:10.1007/978-3-319-12643-2_35.
[11] LABBACI H,MEDJAHED B,BINZAGR F,et al.A deep learning approach for web service interactions[C]//IEEE/WIC/ACM International Conference on Web Intelligence.Leipzig:IEEE Press,2017:848-854.DOI:10.1145/3106426.3106492.
[12] SUK H I,LEE S W,SHEN Dinggang,et al.Latent feature representation with stacked auto-encoder for AD/MCI diagnosis[J].Brain Structure and Function,2015,220(2):841-859.DOI:10.1007/s00429-013-0687-3.
[13] LÜ Yisheng,DUAN Yanjie,KANG Wenwen,et al.Traffic flow prediction with big data: A deep learning approach[J].IEEE Transactions on Intelligent Transportation Systems,2015,16(2):865-873.DOI:10.1109/TITS.2014.2345663.
[14] KAMPER H,ELSNER M,JANSEN A,et al.Unsupervised neural network based feature extraction using weak top-down constraints[C]//IEEE International Conference on Acoustics, Speech and Signal Processing.South Brisbane:IEEE Press,2015:5818-5822.DOI:10.1109/ICASSP.2015.7179087.
[15] MONTAVON G,ORR G,MLLER K R.Neural networks: Tricks of the trade[M].Berlin:Springer,2012:437-478.
[16] 吴琼,陈锻生.多尺度卷积循环神经网络的情感分类技术[J].华侨大学学报(自然科学版),2017,36(6):875-879.DOI:10.11830/ISSN.1000-5013.201606077.

备注/Memo

备注/Memo:
收稿日期: 2017-07-03
通信作者: 张永梅(1967-),女,教授,博士,主要从事图像处理的研究.E-mail:zhangym@ncut.edu.cn.
基金项目: 国家自然科学基金资助项目(61371143); 北方工业大学优势学科科研基金资助项目(XN044); 太原科技大学博士科研启动项目(20162036); 北方工业大学教育教学改革和课程建设研究项目(XN093-002)
更新日期/Last Update: 2018-09-20