您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报 (工学版) ›› 2020, Vol. 50 ›› Issue (3): 58-65.doi: 10.6040/j.issn.1672-3961.0.2019.295

• 机器学习与数据挖掘 • 上一篇    下一篇

基于核极限学习机自编码器的标记分布学习

王一宾1,2(),李田力1,程玉胜1,2,*(),钱坤1   

  1. 1. 安庆师范大学计算机与信息学院,安徽 安庆 246133
    2. 安徽省高校智能感知与计算重点实验室,安徽 安庆 246133
  • 收稿日期:2019-06-10 出版日期:2020-06-01 发布日期:2020-06-16
  • 通讯作者: 程玉胜 E-mail:wangyb07@mail.ustc.edu.cn;chengyshaq@163.com
  • 作者简介:王一宾(1970—),男,安徽安庆人,教授,硕士,主要研究方向为软件安全与多标记学习. E-mail: wangyb07@mail.ustc.edu.cn
  • 基金资助:
    安徽省高校重点自然科学基金资助项目(KJ2017A352);安徽省高校重点实验室基金资助项目(ACAIM160102)

Label distribution learning based on kernel extreme learning machine auto-encoder

Yibin WANG1,2(),Tianli LI1,Yusheng CHENG1,2,*(),Kun QIAN1   

  1. 1. School of Computer and Information, Anqing Normal University, Anqing 246133, Anhui, China
    2. The University Key Laboratory of Intelligent Perception and Computing of Anhui Province, Anqing 246133, Anhui, China
  • Received:2019-06-10 Online:2020-06-01 Published:2020-06-16
  • Contact: Yusheng CHENG E-mail:wangyb07@mail.ustc.edu.cn;chengyshaq@163.com
  • Supported by:
    安徽省高校重点自然科学基金资助项目(KJ2017A352);安徽省高校重点实验室基金资助项目(ACAIM160102)

摘要:

标记分布学习中示例由多个不同重要程度的标记共同标注,而在已有的标记分布学习算法中,大部分均在完备数据集下进行,未考虑数据噪声干扰。针对这一问题,结合自编码器的降噪特性和核极限学习机的稳定性,提出一种基于核极限学习机自编码器的标记分布学习算法。使用核极限学习机自编码器对原始特征空间映射,得到更具鲁棒性的特征表达,构造适应标记分布学习的极限学习机模型作为分类器以提升分类效率及性能。试验结果表明,本文算法较其他对比算法具有一定优势,使用假设检验方法进一步说明所提算法的有效性。

关键词: 标记分布学习, 高斯噪声, 自编码器, 核极限学习机, 鲁棒性

Abstract:

In the label distribution learning framework, the example could be associated with the degree of description of the label. However, most of the algorithms were designed with complete data, and didn′t consider the noise in the data. Therefore, combined the noise reduction characteristics of the auto-encoder and the stability of the kernel extreme learning machine, the Label Distribution Learning Algorithm based on Kernel Extreme Learning Machine with auto-encoder was proposed in this paper. Firstly, we used the auto-encoder in kernel extreme learning machine to map the original feature space to obtain more robust feature representation. Secondly, we constructed the extreme learning machine model that adapted to the label distribution learning as a classifier to improve the classification efficiency and performance. Finally, the experimental results showed the proposed algorithm had certain advantages over other label distribution learning algorithms, and the hypothesis test method further illustrated the effectiveness of the algorithm.

Key words: label distribution learning, Gaussian noise, auto-encoder, kernel extreme learning machine, robustness

中图分类号: 

  • TP181

图1

风景图中的标记分布学习"

图2

AKELM-LDL算法建模过程"

表1

标记分布学习数据集"

数据集 样本数 特征数 标记数
Yeast-alpha 2 465 24 18
Yeast-cdc 2 465 24 15
Yeast-diau 2 465 24 7
Yeast-heat 2 465 24 6
Yeast-spo 2 465 24 6
Yeast-cold 2 465 24 4
Yeast-dtt 2 465 24 4
Yeast-spo5 2 465 24 3
Yeast-elu 2 465 24 14
Human Gene 30 542 36 68
SDU_3DFE 2 500 243 6
Movie 7 755 1 869 5

表2

标记分布评价指标"

评价指标 Chebyshev↓ Clark↓ Canberra↓ Kullback-Leibler↓ Cosine↑ Intersection↑
计算公式 ${\rm{Di}}{{\rm{s}}_1}\left( {D, \hat D} \right) = {\max _j}\left| {{d_j} - {{\hat d}_j}} \right| $ ${\rm{Di}}{{\rm{s}}_2}\left( {D, \hat D} \right) = \sqrt {\sum\limits_{j = 1}^c {\frac{{{{\left( {{d_j} - {{\hat d}_j}} \right)}^2}}}{{{{\left( {{d_j} + {{\hat d}_j}} \right)}^2}}}} } $ ${\rm{Di}}{{\rm{s}}_3}\left( {D, \hat D} \right) = \sum\limits_{j = 1}^c {\frac{{\left| {{d_j} - {{\hat d}_j}} \right|}}{{{d_j} + {{\hat d}_j}}}} $ ${\rm{Di}}{{\rm{s}}_4}\left( {D, \hat D} \right) = \sum\limits_{j = 1}^c {{d_j}\ln \frac{{{d_j}}}{{{{\hat d}_j}}}} $ ${\rm{Si}}{{\rm{m}}_1}\left( {D, \hat D} \right) = \frac{{\sum\limits_{j = 1}^c {{d_j}{{\hat d}_j}} }}{{\sqrt {\sum\limits_{j = 1}^c {{d_j}^2} } \sqrt {\sum\limits_{j = 1}^c {{{\hat d}^2}_j} } }} $ ${\rm{Si}}{{\rm{m}}_2}\left( {D, \hat D} \right) = \sum\limits_{j = 1}^c {\min \left( {{d_j}, {{\hat d}_j}} \right)} $

表3

Chebyshev(↓)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 0.099 6±0.005 1 0.014 6±0.000 4 0.036 7±0.001 3 0.017 0±0.000 5 0.013 5±0.000 3 0.013 4±0.000 1
Yeast-cdc 0.108 7±0.010 1 0.017 5±0.000 4 0.038 8±0.002 1 0.020 0±0.000 5 0.016 3±0.000 4 0.016 1±0.000 1
Yeast-elu 0.112 1±0.006 1 0.017 6±0.000 4 0.038 9±0.001 8 0.020 3±0.000 4 0.016 3±0.000 4 0.016 2±0.000 1
Yeast-diau 0.157 7±0.006 9 0.039 1±0.001 3 0.050 1±0.002 0 0.041 2±0.000 9 0.036 7±0.001 2 0.036 6±0.001 0
Yeast-heat 0.174 1±0.010 5 0.044 7±0.001 7 0.053 0±0.002 7 0.046 6±0.001 1 0.042 4±0.001 4 0.041 4±0.001 1
Yeast-spo 0.175 4±0.009 0 0.062 9±0.002 7 0.067 1±0.003 5 0.061 3±0.001 5 0.058 2±0.002 7 0.057 4±0.001 5
Yeast-cold 0.183 1±0.013 3 0.055 0±0.001 7 0.059 1±0.002 6 0.056 6±0.001 6 0.051 1±0.002 1 0.051 0±0.001 3
Yeast-dtt 0.181 8±0.013 4 0.039 7±0.001 7 0.043 3±0.001 5 0.043 6±0.001 3 0.035 9±0.001 4 0.035 8±0.001 4
Yeast-spo5 0.201 3±0.013 3 0.097 0±0.005 4 0.094 2±0.003 5 0.095 0±0.002 3 0.091 4±0.002 2 0.089 7±0.004 6
Movie 0.201 4±0.002 8 0.124 0±0.002 6 0.139 1±0.003 4 0.147 3±0.002 0 0.126 4±0.003 3 0.114 4±0.001 5
SDU_3DFE 0.138 6±0.003 7 0.127 6±0.003 5 0.142 9±0.006 4 0.134 4±0.005 0 0.104 5±0.002 5 0.133 9±0.003 2
Human Gene 0.195 1±0.034 0 0.065 2±0.001 5 0.063 3±0.001 8 0.054 0±0.002 8 0.053 8±0.002 6 0.053 7±0.002 0
Average 5.911 6 3.250 0 4.666 6 4.000 0 2.000 0 1.166 6

表4

Clark(↓)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 1.172 9±0.039 5 0.230 5±0.004 1 0.734 9±0.031 1 0.261 4±0.007 1 0.210 1±0.003 9 0.209 5±0.006 8
Yeast-cdc 1.080 1±0.061 9 0.235 4±0.004 9 0.595 7±0.036 2 0.257 8±0.006 3 0.216 2±0.005 2 0.213 7±0.004 5
Yeast-elu 1.034 3±0.049 6 0.217 1±0.005 7 0.544 9±0.027 9 0.240 4±0.003 5 0.199 2±0.003 5 0.198 5±0.004 2
Yeast-diau 0.753 1±0.033 3 0.211 0±0.006 2 0.275 3±0.010 4 0.221 5±0.005 2 0.199 1±0.006 4 0.198 3±0.005 4
Yeast-heat 0.683 8±0.038 6 0.193 9±0.007 0 0.232 1±0.013 2 0.201 0±0.004 7 0.184 3±0.005 8 0.179 3±0.004 3
Yeast-spo 0.684 3±0.030 2 0.266 8±0.010 3 0.290 7±0.016 4 0.262 4±0.007 4 0.249 5±0.012 5 0.246 2±0.004 6
Yeast-cold 0.495 4±0.035 1 0.149 7±0.004 7 0.160 9±0.006 9 0.153 2±0.004 9 0.139 7±0.006 2 0.139 1±0.003 1
Yeast-dtt 0.498 6±0.035 8 0.107 5±0.005 1 0.118 3±0.004 2 0.117 2±0.003 9 0.097 8±0.003 9 0.097 7±0.003 6
Yeast-spo5 0.418 0±0.027 8 0.195 8±0.011 3 0.189 5±0.007 3 0.191 4±0.005 1 0.184 4±0.003 6 0.181 2±0.010 4
Movie 0.806 5±0.008 6 0.548 8±0.010 2 0.640 5±0.016 5 0.582 4±0.007 6 0.551 9±0.012 9 0.527 2±0.034 8
SDU_3DFE 0.412 5±0.007 0 0.403 1±0.008 7 0.465 4±0.019 9 0.413 8±0.007 2 0.349 7±0.005 9 0.405 2±0.009 2
Human Gene 4.639 3±0.171 6 2.388 0±0.022 9 3.686 5±0.062 9 2.134 9±0.031 6 2.118 9±0.033 0 2.131 4±0.002 0
Average 5.833 3 3.250 0 4.911 6 3.833 3 1.911 6 1.250 0

表5

Canberra(↓)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 4.193 7±0.149 2 0.753 2±0.014 0 2.420 8±0.095 6 0.861 8±0.023 2 0.682 3±0.014 7 0.680 4±0.024 1
Yeast-cdc 3.541 2±0.225 8 0.712 1±0.014 7 1.798 6±0.108 4 0.783 1±0.015 7 0.647 9±0.017 3 0.640 1±0.017 0
Yeast-elu 3.277 8±0.171 2 0.641 9±0.017 9 1.598 3±0.083 0 0.712 5±0.009 3 0.584 0±0.007 3 0.581 3±0.004 2
Yeast-diau 1.708 6±0.084 0 0.453 9±0.015 1 0.594 4±0.020 7 0.478 7±0.010 1 0.427 0±0.013 5 0.425 7±0.010 8
Yeast-heat 1.444 0±0.082 7 0.390 3±0.013 5 0.467 1±0.023 6 0.404 2±0.009 8 0.367 8±0.010 2 0.357 4±0.008 7
Yeast-spo 1.443 1±0.068 9 0.549 0±0.022 0 0.595 1±0.033 7 0.539 4±0.015 9 0.513 8±0.024 1 0.506 3±0.008 7
Yeast-cold 0.866 9±0.062 0 0.259 3±0.007 2 0.277 3±0.012 3 0.264 9±0.008 5 0.240 6±0.009 3 0.239 7±0.005 4
Yeast-dtt 0.871 9±0.064 9 0.184 7±0.008 1 0.204 1±0.007 6 0.202 7±0.007 2 0.167 9±0.006 6 0.167 8±0.005 3
Yeast-spo5 0.648 8±0.045 0 0.300 3±0.017 0 0.291 1±0.011 2 0.293 9±0.007 7 0.283 0±0.006 4 0.277 9±0.015 2
Movie 1.563 5±0.018 8 1.055 3±0.021 0 1.223 8±0.030 5 1.119 9±0.016 4 1.063 5±0.026 0 0.999 5±0.036 1
SDU_3DFE 0.902 5±0.016 0 0.831 5±0.019 3 0.982 3±0.038 5 0.896 9±0.018 0 0.728 3±0.013 8 0.889 8±0.003 9
HumanGene 33.915 0±1.480 0 16.277 4±0.174 0 25.297 2±0.487 0 14.633 4±0.249 0 14.511 6±0.252 0 14.592 7±0.290 9
Average 5.916 6 3.333 3 4.833 3 3.750 0 1.911 6 1.250 0

表6

Kullback-Leibler(↓)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 0.277 6±0.023 8 0.006 5±0.000 2 0.087 2±0.947 5 0.008 5±0.000 4 0.005 6±0.000 3 0.005 5±0.000 1
Yeast-cdc 0.283 1±0.042 1 0.008 2±0.000 3 0.067 2±0.009 8 0.009 9±0.000 5 0.007 0±0.000 3 0.006 9±0.000 1
Yeast-elu 0.282 5±0.032 8 0.007 3±0.000 4 0.060 8±0.009 2 0.009 1±0.000 3 0.006 2±0.000 2 0.006 1±0.000 1
Yeast-diau 0.269 8±0.027 6 0.014 9±0.000 9 0.026 4±0.002 4 0.015 7±0.000 6 0.012 9±0.000 9 0.012 8±0.000 1
Yeast-heat 0.268 4±0.034 5 0.014 4±0.001 0 0.021 6±0.002 9 0.015 2±0.000 7 0.012 8±0.000 7 0.012 2±0.000 1
Yeast-spo 0.278 8±0.039 8 0.029 1±0.002 3 0.033 2±0.003 9 0.026 8±0.001 5 0.024 6±0.002 3 0.024 0±0.000 1
Yeast-cold 0.217 4±0.035 6 0.013 9±0.001 1 0.016 2±0.001 6 0.014 6±0.001 0 0.012 2±0.001 2 0.012 1±0.000 1
Yeast-dtt 0.226 4±0.039 3 0.007 4±0.000 8 0.008 9±0.000 6 0.008 0±0.000 6 0.006 2±0.000 6 0.006 1±0.000 1
Yeast-spo5 0.206 6±0.042 7 0.034 7±0.003 9 0.031 2±0.002 3 0.031 4±0.001 5 0.029 3±0.001 2 0.028 6±0.002 9
Movie 0.729 7±0.059 0 0.117 7±0.005 1 0.166 4±0.010 7 0.131 7±0.004 6 0.118 9±0.006 0 0.100 3±0.002 7
SDU_3DFE 0.084 9±0.002 9 0.081 8±0.003 9 0.102 3±0.009 6 0.081 9±0.004 0 0.054 1±0.002 2 0.072 6±0.003 3
Human Gene 1.803 5±0.148 7 0.301 9±0.007 2 0.598 4±0.022 1 0.240 6±0.012 1 0.238 6±0.011 2 0.238 7±0.009 8
Average 6.000 0 3.333 3 4.833 3 3.750 0 1.911 6 1.166 6

表7

Cosine(↑)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 0.848 5±0.007 0 0.993 6±0.000 2 0.947 5±0.003 2 0.991 4±0.000 4 0.994 6±0.000 3 0.994 7±0.000 1
Yeast-cdc 0.850 8±0.013 0 0.992 1±0.000 3 0.956 4±0.004 5 0.990 2±0.000 4 0.993 3±0.000 3 0.993 4±0.000 1
Yeast-elu 0.853 1±0.009 2 0.992 9±0.000 4 0.959 7±0.003 9 0.991 0±0.000 3 0.994 0±0.002 0 0.994 1±0.000 1
Yeast-diau 0.863 8±0.007 0 0.986 3±0.000 9 0.977 3±0.001 7 0.985 3±0.000 5 0.988 1±0.007 0 0.988 2±0.000 1
Yeast-heat 0.866 8±0.010 3 0.986 3±0.000 9 0.980 3±0.002 0 0.985 4±0.000 6 0.987 8±0.000 6 0.988 4±0.000 1
Yeast-spo 0.861 1±0.009 3 0.972 7±0.002 2 0.969 5±0.003 2 0.974 7±0.001 3 0.977 0±0.002 0 0.977 5±0.000 1
Yeast-cold 0.893 6±0.009 5 0.986 8±0.000 8 0.984 8±0.001 4 0.986 1±0.000 8 0.988 6±0.001 0 0.988 7±0.000 1
Yeast-dtt 0.895 0±0.009 8 0.992 9±0.000 6 0.991 6±0.000 5 0.991 5±0.000 5 0.994 1±0.000 4 0.994 2±0.000 1
Yeast-spo5 0.898 0±0.010 6 0.969 4±0.003 3 0.972 3±0.001 8 0.972 1±0.001 2 0.974 1±0.000 9 0.974 8±0.002 3
Movie 0.849 5±0.002 2 0.922 4±0.003 2 0.902 8±0.004 5 0.908 1±0.003 0 0.923 5±0.003 7 0.934 2±0.001 5
SDU_3DFE 0.917 9±0.002 8 0.920 2±0.003 5 0.903 7±0.006 9 0.920 3±0.003 6 0.947 0±0.002 2 0.929 6±0.002 9
Human Gene 0.457 0±0.047 7 0.767 4±0.004 0 0.687 6±0.009 8 0.831 6±0.005 5 0.833 3±0.005 3 0.832 1±0.004 3
Average 5.911 6 3.500 0 4.833 3 3.750 0 1.833 3 1.166 6

表8

Intersetion(↑)评价指标结果"

Datasets PT-Bayes AA-KNN AA-BP SA-IIS SA-BFGS AKELM-LDL
Yeast-alpha 0.772 5±0.007 8 0.958 4±0.000 8 0.874 0±0.004 3 0.951 8±0.001 3 0.962 3±0.000 9 0.962 4±0.001 4
Yeast-cdc 0.771 1±0.014 6 0.953 1±0.001 0 0.886 7±0.006 3 0.947 8±0.000 9 0.957 4±0.001 2 0.957 9±0.001 3
Yeast-elu 0.772 7±0.010 9 0.954 7±0.001 3 0.891 7±0.005 2 0.949 1±0.000 7 0.958 8±0.000 5 0.959 0±0.001 1
Yeast-diau 0.769 0±0.010 8 0.937 0±0.002 2 0.917 7±0.002 8 0.933 1±0.001 3 0.940 8±0.001 9 0.941 0±0.001 5
Yeast-heat 0.770 8±0.012 2 0.935 9±0.002 2 0.923 5±0.003 5 0.933 2±0.001 6 0.939 7±0.001 6 0.941 4±0.001 4
Yeast-spo 0.769 1±0.010 8 0.909 5±0.003 7 0.902 2±0.005 3 0.910 9±0.002 6 0.915 4±0.003 7 0.916 6±0.001 4
Yeast-cold 0.796 5±0.013 9 0.936 0±0.001 6 0.931 7±0.003 1 0.934 4±0.002 0 0.940 7±0.002 1 0.940 9±0.001 4
Yeast-dtt 0.796 4±0.014 7 0.954 4±0.001 8 0.949 6±0.001 9 0.949 6±0.001 8 0.958 6±0.001 6 0.958 6±0.001 2
Yeast-spo5 0.798 7±0.013 3 0.903 0±0.005 4 0.905 8±0.003 5 0.905 0±0.002 3 0.908 6±0.002 2 0.910 3±0.004 6
Movie 0.722 6±0.002 8 0.822 4±0.003 8 0.797 6±0.005 1 0.803 2±0.003 3 0.822 1±0.004 6 0.836 4±0.001 6
SDU_3DFE 0.838 8±0.003 2 0.847 9±0.003 8 0.823 6±0.006 7 0.839 4±0.003 7 0.870 9±0.002 7 0.848 8±0.003 9
Human Gene 0.477 9±0.024 3 0.741 7±0.002 7 0.636 2±0.007 8 0.781 4±0.004 0 0.783 4±0.003 9 0.781 6±0.004 1
Average 5.911 6 3.333 3 4.833 3 3.833 3 1.911 6 1.166 6

图3

6种标记分布学习算法结果的雷达图"

1 ZHANG Minling , ZHOU Zhihua . A review on multi-label learning algorithms[J]. IEEE Transactions on Knowledge & Data Engineering, 2014, 26 (8): 1819- 1837.
2 GIBAJA E . A tutorial on multilabel learning[J]. ACM Computing Surveys, 2015, 47 (3): 1- 38.
3 GENG Xin . Label distribution learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2016, 28 (7): 1734- 1748.
doi: 10.1109/TKDE.2016.2545658
4 程玉胜, 陈飞, 王一宾. 基于粗糙集的数据流多标记分布特征选择[J]. 计算机应用, 2018, 38 (11): 3105- 3111.
doi: 10.11772/j.issn.1001-9081.2018041275
CHENG Yusheng , CHEN Fei , WANG Yibin . Feature selection for multi-label distribution learning with streaming data based on rough set[J]. Journal of Computer Application, 2018, 38 (11): 3105- 3111.
doi: 10.11772/j.issn.1001-9081.2018041275
5 季荣姿.标记分布学习及其应用[D].南京:东南大学, 2014.
JI Rongzi. Label distribution learning and its applications[D]. Nanjing: Southeast University, 2014.
6 GENG Xin, HOU Peng. Pre-release prediction of crowd opinion on movies by label distribution learning[C]// International Conference on Artificial Intelligence. Buenos Aires, Argentina: AAAI Press, 2015: 3511-3517.
7 GENG Xin, XIA Yu. Head pose estimation based on multivariate label distribution[C]// Computer Vision and Pattern Recognition. Columbus, America: IEEE, 2014: 1837-1842.
8 王一宾, 田文泉, 程玉胜, 等. 基于核极限学习机的标记分布学习[J]. 计算机工程与应用, 2018, 54 (24): 128- 135.
doi: 10.3778/j.issn.1002-8331.1808-0341
WANG Yibing , TIAN Wenquan , CHENG Yusheng , et al. Label distribution learning based on kernel extreme learning machine[J]. Computer Engineering and Applications, 2018, 54 (24): 128- 135.
doi: 10.3778/j.issn.1002-8331.1808-0341
9 GENG Xin , YIN Chao , ZHOU Zhihua . Facial age estimation by learning from label distributions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35 (10): 2401- 2412.
doi: 10.1109/TPAMI.2013.51
10 ZHOU Ying, XUE Hui, GENG Xin. Emotion distribution recognition from facial expressions[C]// Proceedings of the 23rd Annual ACM Conference on Multimedia. Brisbane, Australia: IEEE, 2015: 1247-1250.
11 LING Miaogen , GENG Xin . Soft video parsing by label distribution learning[J]. Frontiers of Computer Science, 2019, 13 (2): 302- 317.
doi: 10.1007/s11704-018-8015-y
12 XU Miao, ZHOU Zhihua. Incomplete label distribution learning[C]// Twenty-Sixth International Joint Conference on Artificial Intelligence. Melbourne, Australia: IEEE, 2017: 3175-3181.
13 付文博, 孙涛, 梁藉, 等. 深度学习原理及应用综述[J]. 计算机科学, 2018, 45 (增刊1): 24- 28.
FU Wenbo , SUN Tao , LIANG Ji , et al. Review of principle and application of deep learning[J]. Computer Science, 2018, 45 (Suppl.1): 24- 28.
14 唐朝辉, 朱清新, 洪朝群, 等. 基于自编码器及超图学习的多标签特征提取[J]. 自动化学报, 2016, 42 (7): 1014- 1021.
TANG Chaohui , ZHU Qingxin , HONG Chaoqun , et al. Multi-label feature selection with autoencoders and hypergraph learning[J]. Acta Automatica Sinica, 2016, 42 (7): 1014- 1021.
15 杨文元. 多标记学习自编码网络无监督维数约简[J]. 智能系统学报, 2018, 13 (5): 808- 817.
YANG Wenyuan . Multi-label unsupervised dimensionality reduction via autoencoder networks[J]. CAAI Transa-ctions on Intelligent Systems, 2018, 13 (5): 808- 817.
16 LIAN Siming, LIU Jianwei, LU Runkun, et al. Multi-label learning via supervised autoencoder[C]// Proceedings of the 37th China Control Conference. Wuhan, China: IEEE, 2018.
17 HUANG Guangbin , ZHU Qinyu , SIEW Chee-Kheong . Extreme learning machine: theory and applications[J]. Neurocomputing, 2006, 70 (1-3): 489- 501.
doi: 10.1016/j.neucom.2005.12.126
18 WANG Zhiqiong , XIN Junchang , YANG Hongxu , et al. Distributed and weighted extreme learning machine for imbalanced big data learning[J]. Tsinghua Science and Technology, 2017, 22 (2): 160- 173.
doi: 10.23919/TST.2017.7889638
19 耿新, 徐宁, 邵瑞枫. 面向标记分布学习的标记增强[J]. 计算机研究与发展, 2017, 54 (6): 1171- 1184.
GENG Xin , XU Ning , SHAO Ruifeng . Label enhancement for label distribution learning[J]. Journal of Computer Research and Development, 2017, 54 (6): 1171- 1184.
20 LIN Yaojin , LI Yuwen , WANG Chenxi , et al. Attribute reduction for multi-label learning with fuzzy rough set[J]. Knowledge-Based Systems, 2018, 15 (2): 51- 56.
[1] 向润,陈素芬,曾雪强. 基于多重多元回归的人脸年龄估计[J]. 山东大学学报 (工学版), 2019, 49(2): 54-60.
[2] 侯明冬,王印松,田杰. 积分时滞对象的一种内模PID鲁棒控制方法[J]. 山东大学学报(工学版), 2016, 46(5): 64-67.
[3] 张井岗,马文廷,赵志诚. 串级时滞过程的二自由度Smith预估控制[J]. 山东大学学报 (工学版), 2015, 45(5): 43-50.
[4] 葛凯蓉, 常发亮, 董文会. 基于局部敏感直方图的稀疏表达跟踪算法[J]. 山东大学学报(工学版), 2014, 44(5): 14-19.
[5] 谢志华. 一种新的血流建模方法及其在红外人脸识别中的应用[J]. 山东大学学报(工学版), 2013, 43(5): 1-5.
[6] 赵占山1,2, 张静3, 孙连坤1, 丁刚1. 有限时间收敛的滑模自适应控制器设计[J]. 山东大学学报(工学版), 2012, 42(4): 74-78.
[7] 周长辉1, 胡永健2,余绍鹏1. 鲁棒的源扫描仪辨识算法设计[J]. 山东大学学报(工学版), 2011, 41(2): 62-65.
[8] 黄斌. 基于离散趋近律的变结构系统控制[J]. 山东大学学报(工学版), 2011, 41(1): 45-48.
[9] 李岳阳,王士同. 基于鲁棒性神经模糊网络的脉冲噪声滤波算法[J]. 山东大学学报(工学版), 2010, 40(5): 164-170.
[10] 赵永国,贾磊,蔡文剑 . 一种积分过程PID自整定方法[J]. 山东大学学报(工学版), 2008, 38(1): 48-51 .
[11] 解淑英,张承进 . 一种输入饱和受限系统的自适应逆控制[J]. 山东大学学报(工学版), 2006, 36(6): 62-66 .
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 张永花,王安玲,刘福平 . 低频非均匀电磁波在导电界面的反射相角[J]. 山东大学学报(工学版), 2006, 36(2): 22 -25 .
[2] 孔祥臻,刘延俊,王勇,赵秀华 . 气动比例阀的死区补偿与仿真[J]. 山东大学学报(工学版), 2006, 36(1): 99 -102 .
[3] 陈瑞,李红伟,田靖. 磁极数对径向磁轴承承载力的影响[J]. 山东大学学报(工学版), 2018, 48(2): 81 -85 .
[4] 王波,王宁生 . 机电装配体拆卸序列的自动生成及组合优化[J]. 山东大学学报(工学版), 2006, 36(2): 52 -57 .
[5] 李可,刘常春,李同磊 . 一种改进的最大互信息医学图像配准算法[J]. 山东大学学报(工学版), 2006, 36(2): 107 -110 .
[6] 秦通,孙丰荣*,王丽梅,王庆浩,李新彩. 基于极大圆盘引导的形状插值实现三维表面重建[J]. 山东大学学报(工学版), 2010, 40(3): 1 -5 .
[7] 刘文亮,朱维红,陈涤,张泓泉. 基于雷达图像的运动目标形态检测及跟踪技术[J]. 山东大学学报(工学版), 2010, 40(3): 31 -36 .
[8] 张英,郎咏梅,赵玉晓,张鉴达,乔鹏,李善评 . 由EGSB厌氧颗粒污泥培养好氧颗粒污泥的工艺探讨[J]. 山东大学学报(工学版), 2006, 36(4): 56 -59 .
[9] 孙殿柱,朱昌志,李延瑞 . 散乱点云边界特征快速提取算法[J]. 山东大学学报(工学版), 2009, 39(1): 84 -86 .
[10] 赵然杭,陈守煜 . 水资源数量与质量联合评价理论模型研究[J]. 山东大学学报(工学版), 2006, 36(3): 46 -50 .