App. Res. of Comp. Accepted Paper
Vol. 44, 2027 No. 2

Backdoor poisoned sample detection method based on decoupling training and perturbation behavior discrimination

Dong Yaoconga
Xie Lixiab
a. School of Safety Science and Engineering, b. School of Computer Science and Artificial Intelligence, Civil Aviation University of China, Tianjin 300300, China
DOI: 10.19734/j.issn.1001-3695.2026.05.0182

Abstract

Image classification models are vulnerable to data-poisoning backdoor attacks, and existing poisoned sample detection methods still suffer from insufficient stability under stealthy-trigger and weak-trigger scenarios. This paper proposed a backdoor poisoned sample detection method based on decoupled training and perturbation behavior discrimination to address this problem. The method used high-confidence incorrect-label perturbation and Kullback-Leibler (KL) divergence constraint to enhance the discriminative difference between clean and poisoned samples. Then, the method used a Gaussian mixture model to locate suspicious target classes. The method further combined prediction consistency, target-class confidence, and entropy under slight adversarial perturbation to construct a joint discrimination criterion. Experimental results on the CIFAR-10 and GTSRB datasets showed that the proposed method improved the detection rate by 0.28 and 0.34 percentage points, respectively, compared with the confusion training (CT) method. The method reduced the false positive rates of clean samples to below 1.9% on both datasets. The results show that the proposed method can improve the accuracy and stability of backdoor poisoned sample detection.

Foundation Support

国家自然科学基金民航联合研究基金重点项目(U2433205)
国家重点研发计划项目(2025YFB3110500)

Publish Information

DOI: 10.19734/j.issn.1001-3695.2026.05.0182
Publish at: Application Research of Computers Accepted Paper, Vol. 44, 2027 No. 2

Publish History

[2026-10-23] Accepted Paper

Cite This Article

董耀聪, 谢丽霞. 基于解耦训练和扰动行为判别的后门中毒样本检测方法 [J]. 计算机应用研究, 2027, 44 (2). (2026-10-24). https://doi.org/10.19734/j.issn.1001-3695.2026.05.0182. (Dong Yaocong, Xie Lixia. Backdoor poisoned sample detection method based on decoupling training and perturbation behavior discrimination [J]. Application Research of Computers, 2027, 44 (2). (2026-10-24). https://doi.org/10.19734/j.issn.1001-3695.2026.05.0182. )

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
  • 2020-2025 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)