Learning a Label-Noise Robust Logistic Regression: Analysis and Experiments

被引：0
|
作者：
Bootkrajang, Jakramate [1 ]
Kaban, Ata [1 ]
机构：
[1] Univ Birmingham, Sch Comp Sci, Birmingham B15 2TT, W Midlands, England
来源：
INTELLIGENT DATA ENGINEERING AND AUTOMATED LEARNING - IDEAL 2013 | 2013年 / 8206卷
关键词：
label noise; logistic regression; robust learning; gradient ascent optimisation; generalisation error bounds; DISCRIMINANT-ANALYSIS; INITIAL SAMPLES; MICROARRAYS;
D O I：
暂无
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
Label-noise robust logistic regression (rLR) is an extension of logistic regression that includes a model of random mislabelling. This paper attempts a theoretical analysis of rLR. By decomposing and interpreting the gradient of the likelihood objective of rLR as employed in gradient ascent optimisation, we get insights into the ability of the rLR learning algorithm to counteract the negative effect of mislabelling as a result of an intrinsic re-weighting mechanism. We also give an upper-bound on the error of rLR using Rademacher complexities.
引用
收藏
页码：569 / 576
页数：8
相关论文
共 50 条
  • [21] Provably End-to-end Label-noise Learning without Anchor Points
    Li, Xuefeng
    Liu, Tongliang
    Han, Bo
    Niu, Gang
    Sugiyama, Masashi
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 139, 2021, 139
  • [22] Label-Noise Reduction with Support Vector Machines
    Fefilatyev, Sergiy
    Shreve, Matthew
    Kramer, Kurt
    Hall, Lawrence
    Goldgof, Dmitry
    Kasturi, Rangachar
    Daly, Kendra
    Remsen, Andrew
    Bunke, Horst
    [J]. 2012 21ST INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR 2012), 2012, : 3504 - 3508
  • [23] Confidence Scores Make Instance-dependent Label-noise Learning Possible
    Berthon, Antonin
    Han, Bo
    Niu, Gang
    Liu, Tongliang
    Sugiyama, Masashi
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 139, 2021, 139
  • [24] Learning from crowds with robust logistic regression
    Li, Wenbin
    Li, Chaoqun
    Jiang, Liangxiao
    [J]. INFORMATION SCIENCES, 2023, 639
  • [25] CTW: Confident Time-Warping for Time-Series Label-Noise Learning
    Ma, Peitian
    Liu, Zhen
    Zheng, Junhao
    Wang, Linghao
    Ma, Qianli
    [J]. PROCEEDINGS OF THE THIRTY-SECOND INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, IJCAI 2023, 2023, : 4046 - 4054
  • [26] Instance-Dependent Label-Noise Learning with Manifold-Regularized Transition Matrix Estimation
    Cheng, De
    Liu, Tongliang
    Ning, Yixiong
    Wang, Nannan
    Han, Bo
    Niu, Gang
    Gao, Xinbo
    Sugiyama, Masashi
    [J]. 2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022), 2022, : 16609 - 16618
  • [27] Probabilistic Information-Theoretic Discriminant Analysis for Industrial Label-Noise Fault Diagnosis
    Pu, Xiaokun
    Li, Chunguang
    [J]. IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, 2021, 17 (04) : 2664 - 2674
  • [28] Robust Learning of Multi-Label Classifiers under Label Noise
    Kumar, Himanshu
    Manwani, Naresh
    Sastry, P. S.
    [J]. PROCEEDINGS OF THE 7TH ACM IKDD CODS AND 25TH COMAD (CODS-COMAD 2020), 2020, : 90 - 97
  • [29] Urban Noise Analysis Using Multinomial Logistic Regression
    Geraghty, Dermot
    O'Mahony, Margaret
    [J]. JOURNAL OF TRANSPORTATION ENGINEERING, 2016, 142 (06) : 04016020
  • [30] Improving Generalization by Controlling Label-Noise Information in Neural Network Weights
    Harutyunyan, Hrayr
    Reing, Kyle
    Ver Steeg, Greg
    Galstyan, Aram
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 119, 2020, 119
12345