Principled analytic classifier for positive-unlabeled learning via weighted integral probability metric

被引:0
|
作者
Yongchan Kwon
Wonyoung Kim
Masashi Sugiyama
Myunghee Cho Paik
机构
[1] Seoul National University,Department of Statistics
[2] RIKEN,Center for Advanced Intelligence Project
[3] The University of Tokyo,Department of Complexity Science and Engineering, Graduate School of Frontier Sciences
来源
Machine Learning | 2020年 / 109卷
关键词
Positive and unlabeled learning; Integral probability metric; Excess risk bound; Approximation error; Reproducing kernel Hilbert space;
D O I
暂无
中图分类号
学科分类号
摘要
We consider the problem of learning a binary classifier from only positive and unlabeled observations (called PU learning). Recent studies in PU learning have shown superior performance theoretically and empirically. However, most existing algorithms may not be suitable for large-scale datasets because they face repeated computations of a large Gram matrix or require massive hyperparameter optimization. In this paper, we propose a computationally efficient and theoretically grounded PU learning algorithm. The proposed PU learning algorithm produces a closed-form classifier when the hypothesis space is a closed ball in reproducing kernel Hilbert space. In addition, we establish upper bounds of the estimation error and the excess risk. The obtained estimation error bound is sharper than existing results and the derived excess risk bound has an explicit form, which vanishes as sample sizes increase. Finally, we conduct extensive numerical experiments using both synthetic and real datasets, demonstrating improved accuracy, scalability, and robustness of the proposed algorithm.
引用
收藏
页码:513 / 532
页数:19
相关论文
共 50 条
  • [1] Principled analytic classifier for positive-unlabeled learning via weighted integral probability metric
    Kwon, Yongchan
    Kim, Wonyoung
    Sugiyama, Masashi
    Paik, Myunghee Cho
    MACHINE LEARNING, 2020, 109 (03) : 513 - 532
  • [2] Recovering True Classifier Performance in Positive-Unlabeled Learning
    Jain, Shantanu
    White, Martha
    Radivojac, Predrag
    THIRTY-FIRST AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2017, : 2066 - 2072
  • [3] GradPU: Positive-Unlabeled Learning via Gradient Penalty and Positive Upweighting
    Dai, Songmin
    Li, Xiaoqiang
    Zhou, Yue
    Ye, Xichen
    Liu, Tong
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 6, 2023, : 7296 - +
  • [4] Density Estimators for Positive-Unlabeled Learning
    Basile, Teresa M. A.
    Di Mauro, Nicola
    Esposito, Floriana
    Ferilli, Stefano
    Vergari, Antonio
    NEW FRONTIERS IN MINING COMPLEX PATTERNS, NFMCP 2017, 2018, 10785 : 49 - 64
  • [5] Spotting Fake Reviews via Collective Positive-Unlabeled Learning
    Li, Huayi
    Chen, Zhiyuan
    Liu, Bing
    Wei, Xiaokai
    Shao, Jidong
    2014 IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM), 2014, : 899 - 904
  • [6] Generative Adversarial Positive-Unlabeled Learning
    Hou, Ming
    Chaib-draa, Brahim
    Li, Chao
    Zhao, Qibin
    PROCEEDINGS OF THE TWENTY-SEVENTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2018, : 2255 - 2261
  • [7] Positive-Unlabeled Learning in Streaming Networks
    Chang, Shiyu
    Zhang, Yang
    Tang, Jiliang
    Yin, Dawei
    Chang, Yi
    Hasegawa-Johnson, Mark A.
    Huang, Thomas S.
    KDD'16: PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, 2016, : 755 - 764
  • [8] Positive-Unlabeled Learning for Knowledge Distillation
    Ning Jiang
    Jialiang Tang
    Wenxin Yu
    Neural Processing Letters, 2023, 55 : 2613 - 2631
  • [9] Positive-Unlabeled Learning for Knowledge Distillation
    Jiang, Ning
    Tang, Jialiang
    Yu, Wenxin
    NEURAL PROCESSING LETTERS, 2023, 55 (03) : 2613 - 2631
  • [10] A boosting framework for positive-unlabeled learning
    Zhao, Yawen
    Zhang, Mingzhe
    Zhang, Chenhao
    Chen, Weitong
    Ye, Nan
    Xu, Miao
    STATISTICS AND COMPUTING, 2025, 35 (01)