Learning Sound Event Classifiers from Web Audio with Noisy Labels (2019-01-04T00:00:00.000000Z)

TL;DR

Experiments suggest that training with large amounts of noisy data can outperform training with smaller amounts of carefully-labeled data, and it is shown that noise-robust loss functions can be effective in improving performance in presence of corrupted labels.

Abstract

As sound event classification moves towards larger datasets, issues of label noise become inevitable. Web sites can supply large volumes of user-contributed audio and metadata, but inferring labels from this metadata introduces errors due to unreliable inputs, and limitations in the mapping. There is, however, little research into the impact of these errors. To foster the investigation of label noise in sound event classification we present FSDnoisy18k, a dataset containing 42.5 hours of audio across 20 sound classes, including a small amount of manually-labeled data and a larger quantity of real-world noisy data. We characterize the label noise empirically, and provide a CNN baseline system. Experiments suggest that training with large amounts of noisy data can outperform training with smaller amounts of carefully-labeled data. We also show that noise-robust loss functions can be effective in improving performance in presence of corrupted labels.

Authors

D. Ellis

6 papers

Manoj Plakal

4 papers

Eduardo Fonseca

7 papers

TL;DR

Abstract

Authors

References30 items

Environmental Sound Classification Using Deep Convolutional Neural Networks and Data Augmentation

General-purpose Tagging of Freesound Audio with AudioSet Labels: Task Description, Dataset, and Baseline

Data-efficient weakly supervised learning for low-resource audio event detection using deep learning

A Simple Fusion of Deep and Shallow Learning for Acoustic Scene Classification

Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels

A Closer Look at Weak Label Learning for Audio Events

Iterative Learning with Open-set Noisy Labels

Joint Optimization Framework for Learning with Noisy Labels

Audio Set: An ontology and human-labeled dataset for audio events

Robust Loss Functions under Label Noise for Deep Neural Networks

Learning from Noisy Large-Scale Datasets with Minimal Supervision

Audio event and scene recognition: A unified approach using strongly and weakly labeled data

An approach for self-training audio event detectors using web data

Semi-Supervised Active Learning for Sound Classification in Hybrid Learning Environments

Deep Convolutional Neural Networks and Data Augmentation for Environmental Sound Classification

Identity Mappings in Deep Residual Networks

Chime-home: A dataset for sound source recognition in a domestic environment

ESC: Dataset for Environmental Sound Classification

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Adam: A Method for Stochastic Optimization

Training Deep Neural Networks on Noisy Labels with Bootstrapping

A Dataset and Taxonomy for Urban Sound Research

Classification in the Presence of Label Noise: A Survey

Freesound technical demo

Semi-supervised learning helps in sound event classification

Training general-purpose audio tagging networks with noisy labels and iterative self-verification

Audio tagging system using densely connected convolutional networks

DCASE 2018 task 2: iterative training, label smoothing, and background noise normalization for audio event tagging

Audio tagging system for dcase 2018: Focusing on label noise, data augmentation and its efficient learning

Freesound Datasets: A Platform for the Creation of Open Audio Datasets

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names