App. Res. of Comp. Accepted Paper
Vol. 43, 2026 No. 8

Robust fine-tuning method for CNN via nonlinear rule-generated filters

Fan Linlin
Zhang Junna
College of Computer & Information Engineering, Henan Normal University, Xinxiang Henan 453007, China
DOI: 10.19734/j.issn.1001-3695.2025.12.0504

Abstract

Efficient fine-tuning of convolutional neural network (CNN) using local data on resource-constrained Internet of Things devices to provide high-quality visual services has attracted extensive attention. However, existing CNN-oriented fine-tuning studies are limited, and their robustness is insufficient in complex scenarios. To address these issues, this paper proposed a strong robust fine-tuning method for CNN, named Nonlinear rule-Generated filter Fine-tuning (NoGF) . First, a fine-tuning branch with the same structure was constructed for each convolutional layer. A small number of convolutional kernels were randomly generated as seed kernels, and the remaining kernels were generated from the seed kernels through nonlinear rules. Only the seed kernels were updated during fine-tuning, which reduced the number of trainable parameters. Second, a seed kernel grouping strategy was adopted and combined with different nonlinear rules to generate diverse convolutional kernels, which improved feature representation capability and model robustness. Finally, experiments were conducted on five datasets, including CIFAR-10-C and CIFAR-100-C. On the CIFAR-10-C dataset, NoGF improved accuracy by 5.87% compared with full-parameter fine-tuning while reducing the parameter count by 75%. The results show that NoGF achieves higher accuracy than existing fine-tuning methods on multiple datasets, and its robustness is significantly improved.

Foundation Support

国家自然科学基金资助项目(62472147)

Publish Information

DOI: 10.19734/j.issn.1001-3695.2025.12.0504
Publish at: Application Research of Computers Accepted Paper, Vol. 43, 2026 No. 8

Publish History

[2026-04-22] Accepted Paper

Cite This Article

樊琳琳, 张俊娜. 基于非线性规则生成卷积核的CNN强鲁棒微调方法 [J]. 计算机应用研究, 2026, 43 (8). (2026-04-30). https://doi.org/10.19734/j.issn.1001-3695.2025.12.0504. (Fan Linlin, Zhang Junna. Robust fine-tuning method for CNN via nonlinear rule-generated filters [J]. Application Research of Computers, 2026, 43 (8). (2026-04-30). https://doi.org/10.19734/j.issn.1001-3695.2025.12.0504. )

About the Journal

  • Application Research of Computers Monthly Journal
  • Journal ID ISSN 1001-3695
    CN  51-1196/TP

Application Research of Computers, founded in 1984, is an academic journal of computing technology sponsored by Sichuan Institute of Computer Sciences under the Science and Technology Department of Sichuan Province.

Aiming at the urgently needed cutting-edge technology in this discipline, Application Research of Computers reflects the mainstream technology, hot technology and the latest development trend of computer application research at home and abroad in a timely manner. The main contents of the journal include high-level academic papers in this discipline, the latest scientific research results and major application results. The contents of the columns involve new theories of computer discipline, basic computer theory, algorithm theory research, algorithm design and analysis, blockchain technology, system software and software engineering technology, pattern recognition and artificial intelligence, architecture, advanced computing, parallel processing, database technology, computer network and communication technology, information security technology, computer image graphics and its latest hot application technology.

Application Research of Computers has many high-level readers and authors, and its readers are mainly senior and middle-level researchers and engineers engaged in the field of computer science, as well as teachers and students majoring in computer science and related majors in colleges and universities. Over the years, the total citation frequency and Web download rate of Application Research of Computers have been ranked among the top of similar academic journals in this discipline, and the academic papers published are highly popular among the readers for their novelty, academics, foresight, orientation and practicality.

Indexed & Evaluation

  • The Second National Periodical Award 100 Key Journals
  • Double Effect Journal of China Journal Formation
  • the Core Journal of China (Peking University 2023 Edition)
  • the Core Journal for Science
  • Chinese Science Citation Database (CSCD) Source Journals
  • RCCSE Chinese Core Academic Journals
  • Journal of China Computer Federation
  • 2020-2022 The World Journal Clout Index (WJCI) Report of Scientific and Technological Periodicals
  • Full-text Source Journal of China Science and Technology Periodicals Database
  • Source Journal of China Academic Journals Comprehensive Evaluation Database
  • Source Journals of China Academic Journals (CD-ROM Version), China Journal Network
  • 2017-2019 China Outstanding Academic Journals with International Influence (Natural Science and Engineering Technology)
  • Source Journal of Top Academic Papers (F5000) Program of China's Excellent Science and Technology Journals
  • Source Journal of China Engineering Technology Electronic Information Network and Electronic Technology Literature Database
  • Source Journal of British Science Digest (INSPEC)
  • Japan Science and Technology Agency (JST) Source Journal
  • Russian Journal of Abstracts (AJ, VINITI) Source Journals
  • Full-text Journal of EBSCO, USA
  • Cambridge Scientific Abstracts (Natural Sciences) (CSA(NS)) core journals
  • Poland Copernicus Index (IC)
  • Ulrichsweb (USA)