Toward Robustness against Label Noise in Training Deep Discriminative Neural Networks

被引:0
|
作者
Vandat, Arash [1 ]
机构
[1] D Wave Syst Inc, Burnaby, BC, Canada
关键词
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
Collecting large training datasets, annotated with high-quality labels, is costly and time-consuming. This paper proposes a novel framework for training deep convolutional neural networks from noisy labeled datasets that can be obtained cheaply. The problem is formulated using an undirected graphical model that represents the relationship between noisy and clean labels, trained in a semi-supervised setting. In our formulation, the inference over latent clean labels is tractable and is regularized during training using auxiliary sources of information. The proposed model is applied to the image labeling problem and is shown to be effective in labeling unseen images as well as reducing label noise in training on CIFAR-10 and MS COCO datasets.
引用
收藏
页数:10
相关论文
共 50 条
  • [1] Unified Robust Training for Graph Neural Networks Against Label Noise
    Li, Yayong
    Yin, Jie
    Chen, Ling
    [J]. ADVANCES IN KNOWLEDGE DISCOVERY AND DATA MINING, PAKDD 2021, PT I, 2021, 12712 : 528 - 540
  • [2] Analyzing the Noise Robustness of Deep Neural Networks
    Liu, Mengchen
    Liu, Shixia
    Su, Hang
    Cao, Kelei
    Zhu, Jun
    [J]. 2018 IEEE CONFERENCE ON VISUAL ANALYTICS SCIENCE AND TECHNOLOGY (VAST), 2018, : 60 - 71
  • [3] Analyzing the Noise Robustness of Deep Neural Networks
    Cao, Kelei
    Liu, Mengchen
    Su, Hang
    Wu, Jing
    Zhu, Jun
    Liu, Shixia
    [J]. IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, 2021, 27 (07) : 3289 - 3304
  • [4] Batch Normalization and Dropout Regularization in Training Deep Neural Networks with Label Noise
    Rusiecki, Andrzej
    [J]. INTELLIGENT SYSTEMS DESIGN AND APPLICATIONS, ISDA 2021, 2022, 418 : 57 - 66
  • [5] Trimmed Robust Loss Function for Training Deep Neural Networks with Label Noise
    Rusiecki, Andrzej
    [J]. ARTIFICIAL INTELLIGENCEAND SOFT COMPUTING, PT I, 2019, 11508 : 215 - 222
  • [6] Sequence-discriminative training of deep neural networks
    Vesely, Karel
    Ghoshal, Arnab
    Burgett, Lukas
    Povey, Daniel
    [J]. 14TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2013), VOLS 1-5, 2013, : 2344 - 2348
  • [7] Effect of Label Noise on Robustness of Deep Neural Network Object Detectors
    Adhikari, Bishwo
    Peltomaki, Jukka
    Germi, Saeed Bakhshi
    Rahtu, Esa
    Huttunen, Heikki
    [J]. COMPUTER SAFETY, RELIABILITY, AND SECURITY (SAFECOMP 2021), 2021, 12853 : 239 - 250
  • [8] Training Neural Networks with Random Noise Images for Adversarial Robustness
    Park, Ji-Young
    Liu, Lin
    Li, Jiuyong
    Liu, Jixue
    [J]. PROCEEDINGS OF THE 30TH ACM INTERNATIONAL CONFERENCE ON INFORMATION & KNOWLEDGE MANAGEMENT, CIKM 2021, 2021, : 3358 - 3362
  • [9] An efficient test method for noise robustness of deep neural networks
    Yasuda, Muneki
    Sakata, Hironori
    Cho, Seung-Il
    Harada, Tomochika
    Tanaka, Atushi
    Yokoyama, Michio
    [J]. IEICE NONLINEAR THEORY AND ITS APPLICATIONS, 2019, 10 (02): : 221 - 235
  • [10] Achieving Generalizable Robustness of Deep Neural Networks by Stability Training
    Laermann, Jan
    Samek, Wojciech
    Strodthoff, Nils
    [J]. PATTERN RECOGNITION, DAGM GCPR 2019, 2019, 11824 : 360 - 373