Enabling On-Tiny-Device Model Personalization via Gradient Condensing and Alternant Partial Update

On-device training enables the model to adapt to user-specific data by fine-tuning a pre-trained model locally. As embedded devices become ubiquitous, on-device training is increasingly essential since users can benefit from the personalized model without transmitting data and model parameters to th...

Full description

Saved in:
Bibliographic Details
Published in:2025 62nd ACM/IEEE Design Automation Conference (DAC) pp. 1 - 7
Main Authors: Jia, Zhenge, Shi, Yiyang, Bao, Zeyu, Wang, Zirui, Pang, Xin, Liu, Huiguo, Duan, Yu, Shen, Zhaoyan, Zhao, Mengying
Format: Conference Proceeding
Language:English
Published: IEEE 22.06.2025
Subjects:
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract On-device training enables the model to adapt to user-specific data by fine-tuning a pre-trained model locally. As embedded devices become ubiquitous, on-device training is increasingly essential since users can benefit from the personalized model without transmitting data and model parameters to the server. Despite significant efforts toward efficient training, ondevice training still faces a major challenge: The prohibitive cost of multi-layer backpropagation strains the limited resources of tiny devices. In this paper, we propose an algorithm-system cooptimization framework TinyMP that enables self-adaptive on-tiny-device model personalization. To mitigate backpropagation costs, we introduce Gradient Condensing to condense the gradient map structure, significantly reducing the computational complexity and memory consumption of backpropagation while preserving model performance. To further reduce computation overhead, we propose Alternant Partial Update, a mechanism that locally and alternatively selects essential parameters to update without requiring retraining or offline evolutionary search. Our framework is evaluated through extensive experiments using various CNN models (e.g., MobileNetV2, MCUNet) on embedded devices with minimal resources (e.g., OpenMV-H7 with less than 1MB SRAM and 2 MB Flash). Experimental results show that our framework achieves up to 2.4 \times speedup, 80.8% memory saving, and 30.3% accuracy improvement on downstream tasks, outperforming SOTA approaches.
AbstractList On-device training enables the model to adapt to user-specific data by fine-tuning a pre-trained model locally. As embedded devices become ubiquitous, on-device training is increasingly essential since users can benefit from the personalized model without transmitting data and model parameters to the server. Despite significant efforts toward efficient training, ondevice training still faces a major challenge: The prohibitive cost of multi-layer backpropagation strains the limited resources of tiny devices. In this paper, we propose an algorithm-system cooptimization framework TinyMP that enables self-adaptive on-tiny-device model personalization. To mitigate backpropagation costs, we introduce Gradient Condensing to condense the gradient map structure, significantly reducing the computational complexity and memory consumption of backpropagation while preserving model performance. To further reduce computation overhead, we propose Alternant Partial Update, a mechanism that locally and alternatively selects essential parameters to update without requiring retraining or offline evolutionary search. Our framework is evaluated through extensive experiments using various CNN models (e.g., MobileNetV2, MCUNet) on embedded devices with minimal resources (e.g., OpenMV-H7 with less than 1MB SRAM and 2 MB Flash). Experimental results show that our framework achieves up to 2.4 \times speedup, 80.8% memory saving, and 30.3% accuracy improvement on downstream tasks, outperforming SOTA approaches.
Author Wang, Zirui
Liu, Huiguo
Shen, Zhaoyan
Bao, Zeyu
Zhao, Mengying
Jia, Zhenge
Shi, Yiyang
Pang, Xin
Duan, Yu
Author_xml – sequence: 1
  givenname: Zhenge
  surname: Jia
  fullname: Jia, Zhenge
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 2
  givenname: Yiyang
  surname: Shi
  fullname: Shi, Yiyang
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 3
  givenname: Zeyu
  surname: Bao
  fullname: Bao, Zeyu
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 4
  givenname: Zirui
  surname: Wang
  fullname: Wang, Zirui
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 5
  givenname: Xin
  surname: Pang
  fullname: Pang, Xin
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 6
  givenname: Huiguo
  surname: Liu
  fullname: Liu, Huiguo
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 7
  givenname: Yu
  surname: Duan
  fullname: Duan, Yu
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 8
  givenname: Zhaoyan
  surname: Shen
  fullname: Shen, Zhaoyan
  email: shenzhaoyan@sdu.edu.cn
  organization: Shandong University,School of and Computer Science and Technology,China
– sequence: 9
  givenname: Mengying
  surname: Zhao
  fullname: Zhao, Mengying
  organization: Shandong University,School of and Computer Science and Technology,China
BookMark eNo1j81KAzEUhSPoQmvfQCQvMDU_85Msh2lthUq7aNflZnJHAuOdkgmF-vSOqKsD5-M7cB7YLQ2EjD1LsZBS2Jdl3ZTa5HahhCqmSmplVXHD5rayRmtZCC1yc8_cisD1gT74jrJDoGu2xEtokb8PHnu-xzgOBH34ghQG4pcAfB3BB6TEm4E80vgjA3le9wkjwQT2EFOAnh_PHhI-srsO-hHnfzljx9fVodlk2936ram3GcjKpkx7WULnlKuc08aXysjSe6N8hy4XoEozMZl71Uprdde2uZcFCq1bCzlIpWfs6Xc3IOLpHMMnxOvp_7r-BhC6VEc
ContentType Conference Proceeding
DBID 6IE
6IH
CBEJK
RIE
RIO
DOI 10.1109/DAC63849.2025.11132925
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan (POP) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Xplore
IEEE Proceedings Order Plans (POP) 1998-present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
EISBN 9798331503048
EndPage 7
ExternalDocumentID 11132925
Genre orig-research
GroupedDBID 6IE
6IH
CBEJK
RIE
RIO
ID FETCH-LOGICAL-a179t-3d16afb2b7bb38d62816dd82dfeb40a268b2b14d2c1993fcc4d15e033c9a4a123
IEDL.DBID RIE
IngestDate Wed Oct 01 07:05:15 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a179t-3d16afb2b7bb38d62816dd82dfeb40a268b2b14d2c1993fcc4d15e033c9a4a123
PageCount 7
ParticipantIDs ieee_primary_11132925
PublicationCentury 2000
PublicationDate 2025-June-22
PublicationDateYYYYMMDD 2025-06-22
PublicationDate_xml – month: 06
  year: 2025
  text: 2025-June-22
  day: 22
PublicationDecade 2020
PublicationTitle 2025 62nd ACM/IEEE Design Automation Conference (DAC)
PublicationTitleAbbrev DAC
PublicationYear 2025
Publisher IEEE
Publisher_xml – name: IEEE
Score 2.295133
Snippet On-device training enables the model to adapt to user-specific data by fine-tuning a pre-trained model locally. As embedded devices become ubiquitous,...
SourceID ieee
SourceType Publisher
StartPage 1
SubjectTerms Accuracy
Adaptation models
Backpropagation
Computational modeling
Costs
Data models
Performance evaluation
Servers
Strain
Training
Title Enabling On-Tiny-Device Model Personalization via Gradient Condensing and Alternant Partial Update
URI https://ieeexplore.ieee.org/document/11132925
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07b8MgEEZN1KFTWzVV32Lo6sRgG8wY5dFOqYdEyhYBd5YiRU6UOpHy7ws4adWhQzcMBiSO0_H4Pj5CXo20CJKxCFIs3QYFM-dzMnEpy61zzNjaRmxCTib5fK6KI1k9cGEQMYDPsOuT4S4f1nbnj8p6jSw6z1qkJaVoyFpH1i-LVW_YH7jZlHr6Cc-6p59_yaaEqDG-_Gd_V6Tzw7-jxXdkuSZnWN0QM_I0J_dNP6pouqwO0RC9m1MvZ7aixWlV3fAq6X6p6ds2ALpq6toFj1R3lXUFtL8Kx4CuoPAzR6_obON3_h0yG4-mg_foqJAQaedIdZQAE7o03EhjkhwEz5kAyDmUaNJYc5G7MpYCtx6nV1qbAsswThKrdKpd0Lol7Wpd4R2hlmfaWccIrbIUrFSlVqAkYgLaxXRxTzp-gBab5hGMxWlsHv7IfyQX3gweVcX5E2nX2x0-k3O7r5ef25dgui-cD50x
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3NT8IwFG8UTfSkRoz42YPXwdZ163YkfIgRcQdIuJG27y0hIYPgIPG_t-1A48GDt65d26SvL68fv19_hDwpoRFEEHjAMTcbFIyMz4nQpDTTxjF9rSuxCTEaJdNpmu3I6o4Lg4gOfIZNm3R3-bDUG3tU1qpk0Vl0SI4izplf0bV2vN_AT1vddsfMJ24JKCxq7n__JZzi4kb_7J89npP6DwOPZt-x5YIcYHFJVM8Sncw3fS-88bz49LpoHZ1aQbMFzfbr6opZSbdzSZ_XDtJVUtMuWKy6qSwLoO2FOwg0BZmdO3JBJyu796-TSb837gy8nUaCJ40rlV4IQSxzxZRQKkwgZkkQAyQMclTclyxOTFnAgWmL1Mu15hBE6IehTiWXJmxdkVqxLPCaUM0iaeyjYplGHLRIc5lCKhBDkCaqxw1StwM0W1XPYMz2Y3PzR_4jORmM34az4cvo9ZacWpNYjBVjd6RWrjd4T471tpx_rB-cGb8A9gKgeA
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=2025+62nd+ACM%2FIEEE+Design+Automation+Conference+%28DAC%29&rft.atitle=Enabling+On-Tiny-Device+Model+Personalization+via+Gradient+Condensing+and+Alternant+Partial+Update&rft.au=Jia%2C+Zhenge&rft.au=Shi%2C+Yiyang&rft.au=Bao%2C+Zeyu&rft.au=Wang%2C+Zirui&rft.date=2025-06-22&rft.pub=IEEE&rft.spage=1&rft.epage=7&rft_id=info:doi/10.1109%2FDAC63849.2025.11132925&rft.externalDocID=11132925