Addressing Missing Labels in Large-Scale Sound Event Recognition Using a Teacher-Student Framework With Loss Masking

被引:13
|
作者
Fonseca, Eduardo [1 ]
Hershey, Shawn [2 ]
Plakal, Manoj [2 ]
Ellis, Daniel P. W. [2 ]
Jansen, Aren [2 ]
Moore, R. Channing [2 ]
机构
[1] Univ Pompeu Fabra, Mus Technol Grp, Barcelona 08002, Spain
[2] Google Res, New York, NY 10011 USA
关键词
Sound event recognition; label noise; missing labels; teacher-student; loss masking;
D O I
10.1109/LSP.2020.3006378
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
The study of label noise in sound event recognition has recently gained attention with the advent of larger and noisier datasets. This work addresses the problem of missing labels, one of the big weaknesses of large audio datasets, and one of the most conspicuous issues for AudioSet. We propose a simple and model-agnostic method based on a teacher-student framework with loss masking to first identify the most critical missing label candidates, and then ignore their contribution during the learning process. We find that a simple optimisation of the training label set improves recognition performance without additional computation. We discover that most of the improvement comes from ignoring a critical tiny portion of the missing labels. We also show that the damage done by missing labels is larger as the training set gets smaller, yet it can still be observed even when training with massive amounts of audio. We believe these insights can generalize to other large-scale datasets.
引用
收藏
页码:1235 / 1239
页数:5
相关论文
共 14 条
  • [1] Unsupervised Teacher-Student Model for Large-scale Video Retrieval
    Liang, Dong
    Lin, Lanfen
    Wang, Rui
    Shao, Jie
    Wang, Changhu
    Chen, Yen-Wei
    2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW), 2019, : 1863 - 1867
  • [2] Large-Scale Domain Adaptation via Teacher-Student Learning
    Li, Jinyu
    Seltzer, Michael L.
    Wang, Xi
    Zhao, Rui
    Gong, Yifan
    18TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2017), VOLS 1-6: SITUATED INTERACTION, 2017, : 2386 - 2390
  • [3] Exploiting Large-Scale Teacher-Student Training for On-Device Acoustic Models
    Liu, Jing
    Swaminathan, Rupak Vignesh
    Parthasarathi, Sree Hari Krishnan
    Lyu, Chunchuan
    Mouchtaris, Athanasios
    Kunzmann, Siegfried
    TEXT, SPEECH, AND DIALOGUE, TSD 2021, 2021, 12848 : 413 - 424
  • [4] FACILITATING TEACHER-STUDENT COMMUNICATION AND INTERACTION IN LARGE-SCALE LECTURES WITH SMARTPHONES AND RWTHAPP
    Politze, M.
    Decker, B.
    Schaffert, S.
    Kueppers, B.
    EDULEARN15: 7TH INTERNATIONAL CONFERENCE ON EDUCATION AND NEW LEARNING TECHNOLOGIES, 2015, : 4820 - 4828
  • [5] Teacher-Student Framework for Polyphonic Semi-supervised Sound Event Detection: Survey and Empirical Analysis
    Diffallah, Zhor
    Ykhlef, Hadjer
    Bouarfa, Hafida
    ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2024, 15 (05)
  • [6] Negative teacher-student and student-student relationships are associated with school dropout: Evidence from a large-scale longitudinal study in Chile
    Contreras, Dante
    Gonzalez, Luis
    Lascar, Samuel
    Lopez, Veronica
    INTERNATIONAL JOURNAL OF EDUCATIONAL DEVELOPMENT, 2022, 91
  • [7] Large Dataset Generation of Synchronized Music Audio and Lyrics at Scale using Teacher-Student Paradigm
    Chivriga, Cristian
    Roy, Rinita
    INTERSPEECH 2023, 2023, : 1473 - 1477
  • [8] A FRAMEWORK FOR TEACHER BACKGROUND QUESTIONNAIRES OF LARGE-SCALE ASSESSMENTS WITH A FOCUS ON TEACHING PRACTICES THAT INFLUENCE STUDENT ACHIEVEMENT IN MATHEMATICS
    Simon, Marielle
    Sarwar, Gul Shahzad
    van Barneveld, Christina
    Zerpa, Carlos
    5TH INTERNATIONAL CONFERENCE OF EDUCATION, RESEARCH AND INNOVATION (ICERI 2012), 2012, : 2809 - 2815
  • [9] Large scale continuous visual event recognition using max-margin Hough transformation framework
    Chakraborty, Bhaskar
    Gonzalez, Jordi
    Roca, F. Xavier
    COMPUTER VISION AND IMAGE UNDERSTANDING, 2013, 117 (10) : 1356 - 1368
  • [10] A large-scale sensor missing data imputation framework for dams using deep learning and transfer learning strategy
    Li, Yangtao
    Bao, Tengfei
    Chen, Hao
    Zhang, Kang
    Shu, Xiaosong
    Chen, Zexun
    Hu, Yuhan
    MEASUREMENT, 2021, 178