DAP: A dataset-agnostic predictor of neural network performance

<p>Training a deep neural network on a large dataset to convergence is a time-demanding task. This task often must be repeated many times, especially when developing a new deep learning algorithm or performing a neural architecture search. This problem can be mitigated if a deep neural network...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Sui Paul Ang (18460605) (author)
مؤلفون آخرون: Soan T.M. Duong (19256503) (author), Son Lam Phung (18460602) (author), Abdesselam Bouzerdoum (17900021) (author)
منشور في: 2024
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513509806571520
author Sui Paul Ang (18460605)
author2 Soan T.M. Duong (19256503)
Son Lam Phung (18460602)
Abdesselam Bouzerdoum (17900021)
author2_role author
author
author
author_facet Sui Paul Ang (18460605)
Soan T.M. Duong (19256503)
Son Lam Phung (18460602)
Abdesselam Bouzerdoum (17900021)
author_role author
dc.creator.none.fl_str_mv Sui Paul Ang (18460605)
Soan T.M. Duong (19256503)
Son Lam Phung (18460602)
Abdesselam Bouzerdoum (17900021)
dc.date.none.fl_str_mv 2024-03-18T09:00:00Z
dc.identifier.none.fl_str_mv 10.1016/j.neucom.2024.127544
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/DAP_A_dataset-agnostic_predictor_of_neural_network_performance/26404033
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Neural network performance predictor
Deep learning
Dataset-agnostic
Neural architecture search
AutoML
dc.title.none.fl_str_mv DAP: A dataset-agnostic predictor of neural network performance
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p>Training a deep neural network on a large dataset to convergence is a time-demanding task. This task often must be repeated many times, especially when developing a new deep learning algorithm or performing a neural architecture search. This problem can be mitigated if a deep neural network’s performance can be estimated without actually training it. In this work, we investigate the feasibility of two tasks: (i) predicting a deep neural network’s performance accurately given only its architectural descriptor, and (ii) generalizing the predictor across different datasets without re-training. To this end, we propose a dataset-agnostic regression framework that uses a novel dual-LSTM model and a new dataset difficulty feature. The experimental results show that both tasks above are indeed feasible, and the proposed method outperforms the existing techniques in all experimental cases. Additionally, we also demonstrate several practical use-cases of the proposed predictor.</p><h2>Other Information</h2> <p> Published in: Neurocomputing<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.neucom.2024.127544" target="_blank">https://dx.doi.org/10.1016/j.neucom.2024.127544</a></p>
eu_rights_str_mv openAccess
id Manara2_a6fe8238ed8c29a37b21730be0a029df
identifier_str_mv 10.1016/j.neucom.2024.127544
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/26404033
publishDate 2024
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling DAP: A dataset-agnostic predictor of neural network performanceSui Paul Ang (18460605)Soan T.M. Duong (19256503)Son Lam Phung (18460602)Abdesselam Bouzerdoum (17900021)Information and computing sciencesArtificial intelligenceData management and data scienceMachine learningNeural network performance predictorDeep learningDataset-agnosticNeural architecture searchAutoML<p>Training a deep neural network on a large dataset to convergence is a time-demanding task. This task often must be repeated many times, especially when developing a new deep learning algorithm or performing a neural architecture search. This problem can be mitigated if a deep neural network’s performance can be estimated without actually training it. In this work, we investigate the feasibility of two tasks: (i) predicting a deep neural network’s performance accurately given only its architectural descriptor, and (ii) generalizing the predictor across different datasets without re-training. To this end, we propose a dataset-agnostic regression framework that uses a novel dual-LSTM model and a new dataset difficulty feature. The experimental results show that both tasks above are indeed feasible, and the proposed method outperforms the existing techniques in all experimental cases. Additionally, we also demonstrate several practical use-cases of the proposed predictor.</p><h2>Other Information</h2> <p> Published in: Neurocomputing<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.neucom.2024.127544" target="_blank">https://dx.doi.org/10.1016/j.neucom.2024.127544</a></p>2024-03-18T09:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1016/j.neucom.2024.127544https://figshare.com/articles/journal_contribution/DAP_A_dataset-agnostic_predictor_of_neural_network_performance/26404033CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/264040332024-03-18T09:00:00Z
spellingShingle DAP: A dataset-agnostic predictor of neural network performance
Sui Paul Ang (18460605)
Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Neural network performance predictor
Deep learning
Dataset-agnostic
Neural architecture search
AutoML
status_str publishedVersion
title DAP: A dataset-agnostic predictor of neural network performance
title_full DAP: A dataset-agnostic predictor of neural network performance
title_fullStr DAP: A dataset-agnostic predictor of neural network performance
title_full_unstemmed DAP: A dataset-agnostic predictor of neural network performance
title_short DAP: A dataset-agnostic predictor of neural network performance
title_sort DAP: A dataset-agnostic predictor of neural network performance
topic Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Neural network performance predictor
Deep learning
Dataset-agnostic
Neural architecture search
AutoML