Leverage Score Sampling for Complete Mode Coverage in Generative Adversarial Networks

被引:0
|
作者
Schreurs, Joachim [1 ]
De Meulemeester, Hannes [1 ]
Fanuel, Michael [2 ]
De Moor, Bart [1 ]
Suykens, Johan A. K. [1 ]
机构
[1] Katholieke Univ Leuven, ESAT STADIUS, Kasteelpk Arenberg 10, B-3001 Leuven, Belgium
[2] Univ Lille, CNRS, Cent Lille, UMR 9189,CRIStAL, F-59000 Lille, France
基金
欧洲研究理事会;
关键词
GANs; Leverage score sampling; Complete mode coverage;
D O I
10.1007/978-3-030-95470-3_35
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Commonly, machine learning models minimize an empirical expectation. As a result, the trained models typically perform well for the majority of the data but the performance may deteriorate in less dense regions of the dataset. This issue also arises in generative modeling. A generative model may overlook underrepresented modes that are less frequent in the empirical data distribution. This problem is known as complete mode coverage. We propose a sampling procedure based on ridge leverage scores which significantly improves mode coverage when compared to standard methods and can easily be combined with any GAN. Ridge leverage scores are computed by using an explicit feature map, associated with the next-to-last layer of a GAN discriminator or of a pre-trained network, or by using an implicit feature map corresponding to a Gaussian kernel. Multiple evaluations against recent approaches of complete mode coverage show a clear improvement when using the proposed sampling strategy.
引用
收藏
页码:466 / 480
页数:15
相关论文
共 50 条
  • [1] Reparameterized Sampling for Generative Adversarial Networks
    Wang, Yifei
    Wang, Yisen
    Yang, Jiansheng
    Lin, Zhouchen
    MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, ECML PKDD 2021: RESEARCH TRACK, PT III, 2021, 12977 : 494 - 509
  • [2] Collaborative Sampling in Generative Adversarial Networks
    Liu, Yuejiang
    Kothari, Parth
    Alahi, Alexandre
    THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 4948 - 4956
  • [3] On Mode Collapse in Generative Adversarial Networks
    Zhang, Kaifeng
    ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING - ICANN 2021, PT II, 2021, 12892 : 563 - 574
  • [4] Generalized Leverage Score Sampling for Neural Networks
    Lee, Jason D.
    Shen, Ruoqi
    Song, Zhao
    Wang, Mengdi
    Yu, Zheng
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 33, NEURIPS 2020, 2020, 33
  • [5] Score-Guided Generative Adversarial Networks
    Lee, Minhyeok
    Seok, Junhee
    AXIOMS, 2022, 11 (12)
  • [6] Geometry Score: A Method For Comparing Generative Adversarial Networks
    Khrulkov, Valentin
    Oseledets, Ivan
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 80, 2018, 80
  • [7] Time series (re)sampling using Generative Adversarial Networks?
    Dahl, Christian M.
    Sorensen, Emil N.
    NEURAL NETWORKS, 2022, 156 : 95 - 107
  • [8] Time series (re)sampling using Generative Adversarial Networks
    Dahl, Christian M.
    Sørensen, Emil N.
    Neural Networks, 2022, 156 : 95 - 107
  • [9] Sky Image forecasting with Generative Adversarial Networks for cloud coverage prediction
    Andrianakos, George
    Tsourounis, Dimitrios
    Oikonomou, Spiros
    Kastaniotis, Dimitris
    Economou, George
    Kazantzidis, Andreas
    2019 10TH INTERNATIONAL CONFERENCE ON INFORMATION, INTELLIGENCE, SYSTEMS AND APPLICATIONS (IISA), 2019, : 28 - 34
  • [10] Distribution Fitting for Combating Mode Collapse in Generative Adversarial Networks
    Gong, Yanxiang
    Xie, Zhiwei
    Duan, Guozhen
    Ma, Zheng
    Xie, Mei
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2024, 35 (12) : 18251 - 18262