An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data
With increasing concern for marine species extinction, a massive effort has been made to conserve, prevent, and search for a sustainable solution. However, data labeling is a labor-heavy and time-consuming work, resulting in limited annotated acoustic data. What's more, a majority of labeled ac...
Main Author: | |
---|---|
Other Authors: | , , , , , , , |
Language: | English |
Published: |
2020
|
Subjects: | |
Online Access: | http://hdl.handle.net/10222/79390 |
id |
ftdalhouse:oai:DalSpace.library.dal.ca:10222/79390 |
---|---|
record_format |
openpolar |
spelling |
ftdalhouse:oai:DalSpace.library.dal.ca:10222/79390 2023-05-15T17:28:04+02:00 An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data Liu, Xuhui Faculty of Computer Science Master of Computer Science Dr. David Barclay Michael McAllister Dr. Sageev Oore Dr. Stan Matwin Dr. Oliver Kirsebom Not Applicable 2020-06-11T12:25:35Z http://hdl.handle.net/10222/79390 en eng http://hdl.handle.net/10222/79390 Deep Learning Imbalanced Data North Atlantic Right Whale (NARW) Transfer Learning 2020 ftdalhouse 2022-03-06T00:10:55Z With increasing concern for marine species extinction, a massive effort has been made to conserve, prevent, and search for a sustainable solution. However, data labeling is a labor-heavy and time-consuming work, resulting in limited annotated acoustic data. What's more, a majority of labeled acoustic data are background noise. Both issues together raise interests in searching for solutions on how to effectively train a reliable classification model. We simulate different degrees of data compositions to study the impact of data scarcity and class imbalance on the North Atlantic Right Whale (NARW) acoustic data. In the meantime, we explore two types of supervised deep learning approaches: metric-based classifiers and cross-entropy based classifiers. The empirical results show that our classifiers trained with fewer NARW acoustic data have comparable performance to the-state-of-art classifiers trained with a larger amount of acoustic data. Other/Unknown Material North Atlantic North Atlantic right whale Dalhousie University: DalSpace Institutional Repository |
institution |
Open Polar |
collection |
Dalhousie University: DalSpace Institutional Repository |
op_collection_id |
ftdalhouse |
language |
English |
topic |
Deep Learning Imbalanced Data North Atlantic Right Whale (NARW) Transfer Learning |
spellingShingle |
Deep Learning Imbalanced Data North Atlantic Right Whale (NARW) Transfer Learning Liu, Xuhui An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
topic_facet |
Deep Learning Imbalanced Data North Atlantic Right Whale (NARW) Transfer Learning |
description |
With increasing concern for marine species extinction, a massive effort has been made to conserve, prevent, and search for a sustainable solution. However, data labeling is a labor-heavy and time-consuming work, resulting in limited annotated acoustic data. What's more, a majority of labeled acoustic data are background noise. Both issues together raise interests in searching for solutions on how to effectively train a reliable classification model. We simulate different degrees of data compositions to study the impact of data scarcity and class imbalance on the North Atlantic Right Whale (NARW) acoustic data. In the meantime, we explore two types of supervised deep learning approaches: metric-based classifiers and cross-entropy based classifiers. The empirical results show that our classifiers trained with fewer NARW acoustic data have comparable performance to the-state-of-art classifiers trained with a larger amount of acoustic data. |
author2 |
Faculty of Computer Science Master of Computer Science Dr. David Barclay Michael McAllister Dr. Sageev Oore Dr. Stan Matwin Dr. Oliver Kirsebom Not Applicable |
author |
Liu, Xuhui |
author_facet |
Liu, Xuhui |
author_sort |
Liu, Xuhui |
title |
An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
title_short |
An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
title_full |
An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
title_fullStr |
An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
title_full_unstemmed |
An Empirical Analysis of Cross-entropy Based and Metric-based Methods on North Atlantic Right Whale Acoustic Data |
title_sort |
empirical analysis of cross-entropy based and metric-based methods on north atlantic right whale acoustic data |
publishDate |
2020 |
url |
http://hdl.handle.net/10222/79390 |
genre |
North Atlantic North Atlantic right whale |
genre_facet |
North Atlantic North Atlantic right whale |
op_relation |
http://hdl.handle.net/10222/79390 |
_version_ |
1766120532558741504 |