Efficient multi-objective neural architecture search framework via policy gradient algorithm

<p>Differentiable architecture search plays a prominent role in Neural Architecture Search (NAS) and exhibits preferable efficiency than traditional heuristic NAS methods, including those based on evolutionary algorithms (EA) and reinforcement learning (RL). However, differentiable NAS methods...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Bo Lyu (16522643) (author)
مؤلفون آخرون: Yin Yang (35103) (author), Yuting Cao (4231810) (author), Pengcheng Wang (400334) (author), Jian Zhu (126278) (author), Jingfei Chang (19325611) (author), Shiping Wen (7168688) (author)
منشور في: 2024
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513509741559808
author Bo Lyu (16522643)
author2 Yin Yang (35103)
Yuting Cao (4231810)
Pengcheng Wang (400334)
Jian Zhu (126278)
Jingfei Chang (19325611)
Shiping Wen (7168688)
author2_role author
author
author
author
author
author
author_facet Bo Lyu (16522643)
Yin Yang (35103)
Yuting Cao (4231810)
Pengcheng Wang (400334)
Jian Zhu (126278)
Jingfei Chang (19325611)
Shiping Wen (7168688)
author_role author
dc.creator.none.fl_str_mv Bo Lyu (16522643)
Yin Yang (35103)
Yuting Cao (4231810)
Pengcheng Wang (400334)
Jian Zhu (126278)
Jingfei Chang (19325611)
Shiping Wen (7168688)
dc.date.none.fl_str_mv 2024-01-26T12:00:00Z
dc.identifier.none.fl_str_mv 10.1016/j.ins.2024.120186
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Efficient_multi-objective_neural_architecture_search_framework_via_policy_gradient_algorithm/26490850
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
Machine learning
Neural architecture search
Reinforcement learning
Non-differentiable
Supernetwork
dc.title.none.fl_str_mv Efficient multi-objective neural architecture search framework via policy gradient algorithm
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p>Differentiable architecture search plays a prominent role in Neural Architecture Search (NAS) and exhibits preferable efficiency than traditional heuristic NAS methods, including those based on evolutionary algorithms (EA) and reinforcement learning (RL). However, differentiable NAS methods encounter challenges when dealing with non-differentiable objectives like energy efficiency, resource constraints, and other non-differentiable metrics, especially under multi-objective search scenarios. While the multi-objective NAS research addresses these challenges, the individual training required for each candidate architecture demands significant computational resources. To bridge this gap, this work combines the efficiency of the differentiable NAS with metrics compatibility in multi-objective NAS. The architectures are discretely sampled by the architecture parameter α within the differentiable NAS framework, and α are directly optimised by the policy gradient algorithm. This approach eliminates the need for a sampling controller to be learned and enables the encompassment of non-differentiable metrics. We provide an efficient NAS framework that can be readily customized to address real-world multi-objective NAS (MNAS) scenarios, encompassing factors such as resource limitations and platform specialization. Notably, compared with other multi-objective NAS methods, our NAS framework effectively decreases the computational burden (accounting for just 1/6 of the NSGA-Net). This search framework is also compatible with the other efficiency and performance improvement strategies under the differentiable NAS framework.</p><h2>Other Information</h2> <p> Published in: Information Sciences<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.ins.2024.120186" target="_blank">https://dx.doi.org/10.1016/j.ins.2024.120186</a></p>
eu_rights_str_mv openAccess
id Manara2_6323982993befe7d8dcf86209faa8f8b
identifier_str_mv 10.1016/j.ins.2024.120186
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/26490850
publishDate 2024
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Efficient multi-objective neural architecture search framework via policy gradient algorithmBo Lyu (16522643)Yin Yang (35103)Yuting Cao (4231810)Pengcheng Wang (400334)Jian Zhu (126278)Jingfei Chang (19325611)Shiping Wen (7168688)Information and computing sciencesArtificial intelligenceMachine learningNeural architecture searchReinforcement learningNon-differentiableSupernetwork<p>Differentiable architecture search plays a prominent role in Neural Architecture Search (NAS) and exhibits preferable efficiency than traditional heuristic NAS methods, including those based on evolutionary algorithms (EA) and reinforcement learning (RL). However, differentiable NAS methods encounter challenges when dealing with non-differentiable objectives like energy efficiency, resource constraints, and other non-differentiable metrics, especially under multi-objective search scenarios. While the multi-objective NAS research addresses these challenges, the individual training required for each candidate architecture demands significant computational resources. To bridge this gap, this work combines the efficiency of the differentiable NAS with metrics compatibility in multi-objective NAS. The architectures are discretely sampled by the architecture parameter α within the differentiable NAS framework, and α are directly optimised by the policy gradient algorithm. This approach eliminates the need for a sampling controller to be learned and enables the encompassment of non-differentiable metrics. We provide an efficient NAS framework that can be readily customized to address real-world multi-objective NAS (MNAS) scenarios, encompassing factors such as resource limitations and platform specialization. Notably, compared with other multi-objective NAS methods, our NAS framework effectively decreases the computational burden (accounting for just 1/6 of the NSGA-Net). This search framework is also compatible with the other efficiency and performance improvement strategies under the differentiable NAS framework.</p><h2>Other Information</h2> <p> Published in: Information Sciences<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.ins.2024.120186" target="_blank">https://dx.doi.org/10.1016/j.ins.2024.120186</a></p>2024-01-26T12:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1016/j.ins.2024.120186https://figshare.com/articles/journal_contribution/Efficient_multi-objective_neural_architecture_search_framework_via_policy_gradient_algorithm/26490850CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/264908502024-01-26T12:00:00Z
spellingShingle Efficient multi-objective neural architecture search framework via policy gradient algorithm
Bo Lyu (16522643)
Information and computing sciences
Artificial intelligence
Machine learning
Neural architecture search
Reinforcement learning
Non-differentiable
Supernetwork
status_str publishedVersion
title Efficient multi-objective neural architecture search framework via policy gradient algorithm
title_full Efficient multi-objective neural architecture search framework via policy gradient algorithm
title_fullStr Efficient multi-objective neural architecture search framework via policy gradient algorithm
title_full_unstemmed Efficient multi-objective neural architecture search framework via policy gradient algorithm
title_short Efficient multi-objective neural architecture search framework via policy gradient algorithm
title_sort Efficient multi-objective neural architecture search framework via policy gradient algorithm
topic Information and computing sciences
Artificial intelligence
Machine learning
Neural architecture search
Reinforcement learning
Non-differentiable
Supernetwork