Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

From Weak to Strong Sound Event Labels using Adaptive Change-Point Detection and Active Learning

Martinsson, John LU ; Mogren, Olof ; Sandsten, Maria LU and Virtanen, Tuomas (2024) 32nd European Signal Processing Conference
(EUSIPCO 2024) p.902-906
Abstract
We propose an adaptive change point detection method (A-CPD) for machine guided weak label annotation of audio recording segments. The goal is to maximize the amount of information gained about the temporal activations of the target sounds. For each unlabeled audio recording, we use a prediction model to derive a probability curve used to guide annotation. The prediction model is initially pre-trained on available annotated sound event data with classes that are disjoint from the classes in the unlabeled dataset. The prediction model then gradually adapts to the annotations provided by the annotator in an active learning loop. We derive query segments to guide the weak label annotator towards strong labels, using change point detection on... (More)
We propose an adaptive change point detection method (A-CPD) for machine guided weak label annotation of audio recording segments. The goal is to maximize the amount of information gained about the temporal activations of the target sounds. For each unlabeled audio recording, we use a prediction model to derive a probability curve used to guide annotation. The prediction model is initially pre-trained on available annotated sound event data with classes that are disjoint from the classes in the unlabeled dataset. The prediction model then gradually adapts to the annotations provided by the annotator in an active learning loop. We derive query segments to guide the weak label annotator towards strong labels, using change point detection on these probabilities. We show that it is possible to derive strong labels of high quality with a limited annotation budget, and show favorable results for A-CPD when compared to two baseline query segment strategies. (Less)
Please use this url to cite or link to this publication:
author
; ; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Active learning, Annotation, Sound event detection, Deep learning
host publication
32nd European Signal Processing Conference (EUSIPCO), proceedings of
pages
902 - 906
conference name
32nd European Signal Processing Conference<br/>(EUSIPCO 2024)
conference location
Lyon, France
conference dates
2024-08-26 - 2024-08-30
external identifiers
  • scopus:85208422384
ISBN
978-9-4645-9361-7
DOI
10.23919/eusipco63174.2024.10715098
language
English
LU publication?
yes
id
7f8469f4-d3d3-435d-8760-8d7a6d8df6c4
alternative location
https://eurasip.org/open-library/
date added to LUP
2024-09-20 08:44:45
date last changed
2025-04-04 14:55:47
@inproceedings{7f8469f4-d3d3-435d-8760-8d7a6d8df6c4,
  abstract     = {{We propose an adaptive change point detection method (A-CPD) for machine guided weak label annotation of audio recording segments. The goal is to maximize the amount of information gained about the temporal activations of the target sounds. For each unlabeled audio recording, we use a prediction model to derive a probability curve used to guide annotation. The prediction model is initially pre-trained on available annotated sound event data with classes that are disjoint from the classes in the unlabeled dataset. The prediction model then gradually adapts to the annotations provided by the annotator in an active learning loop. We derive query segments to guide the weak label annotator towards strong labels, using change point detection on these probabilities. We show that it is possible to derive strong labels of high quality with a limited annotation budget, and show favorable results for A-CPD when compared to two baseline query segment strategies.}},
  author       = {{Martinsson, John and Mogren, Olof and Sandsten, Maria and Virtanen, Tuomas}},
  booktitle    = {{32nd European Signal Processing Conference (EUSIPCO), proceedings of}},
  isbn         = {{978-9-4645-9361-7}},
  keywords     = {{Active learning; Annotation; Sound event detection; Deep learning}},
  language     = {{eng}},
  month        = {{08}},
  pages        = {{902--906}},
  title        = {{From Weak to Strong Sound Event Labels using Adaptive Change-Point Detection and Active Learning}},
  url          = {{http://dx.doi.org/10.23919/eusipco63174.2024.10715098}},
  doi          = {{10.23919/eusipco63174.2024.10715098}},
  year         = {{2024}},
}