• Home
  • Search
  • Deep Neural Backdoor in Semi-Supervised Learning: Threats and Countermeasures
  • Cite Icon44
  • https://doi.org/10.1109/tifs.2021.3116431Copy DOI Icon

Deep Neural Backdoor in Semi-Supervised Learning: Threats and Countermeasures

Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Semi-Supervised Learning (SSL) is a powerful derivative for humans to discover the hidden knowledge, and will be a great substitute for data taggers. Although the availability of unlabeled data rises up a huge passion to SSL, the untrustness of unlabeled data leads to many unknown security risks. In this paper, we first identify an insidious backdoor threat of SSL where unlabeled training data are poisoned by backdoor methods migrated from supervised settings. Then, to further exploit this threat, a Deep Neural Backdoor (DeNeB) scheme is proposed, which requires less data poisoning budgets and produces stronger backdoor effectiveness. By poisoning a fraction of our unlabeled training data, the DeNeB achieves the illegal manipulation on the trained model without modifying the training process. Finally, an efficient detection-and-purification defense (DePuD) framework is proposed to thwart the proposed scheme. In DePuD, we construct a deep detector to locate trigger patterns in the unlabeled training data, and perform secured SSL training with purified unlabeled data where the detected trigger patterns are obfuscated. Extensive experiments based on benchmark datasets are performed to demonstrate the huge threatening of DeNeB and the effectiveness of DePuD. To the best of our knowledge, this is the first work to achieve the backdoor and its defense in semi-supervised learning.

Similar Papers
  • Research Article
  • Citations20

SEMI-SUPERVISED SEQUENCE CLASSIFICATION WITH HMMs

  • Mar 01, 2005
  • International Journal of Pattern Recognition and Artificial Intelligence
  • Shi Zhong
  • Conference Article
  • Citations26

SILU: Strategy Involving Large-scale Unlabeled Logs for Improving Malware Detector

  • Jul 01, 2020
  • Taishi Nishiyama +3
  • Book Chapter
  • Citations12

Semi-Supervised Learning

  • Jan 01, 2009
  • Tobias Scheffer
  • Conference Article
  • Citations31

Incorporating Semi-Supervised and Positive-Unlabeled Learning for Boosting Full Reference Image Quality Assessment

  • Jun 01, 2022
  • Yue Cao +4
  • Research Article
  • Citations3

Active Labeling Aided Semi-Supervised Safety Assessment With Task-Related Unknown Scenarios

  • Dec 01, 2024
  • IEEE Transactions on Reliability
  • Chang Liu +4
  • Conference Article
  • Citations2

Benchmarking the semi-supervised naïve Bayes classifier

  • Jul 01, 2015
  • Awat A Saeed +2
  • Research Article
  • Citations216

Driver Distraction Detection Using Semi-Supervised Machine Learning

  • Apr 01, 2016
  • IEEE Transactions on Intelligent Transportation Systems
  • Tianchi Liu +4
  • Research Article
  • Citations8

Geostatistical semi-supervised learning for spatial prediction

  • Dec 01, 2022
  • Artificial Intelligence in Geosciences
  • Francky Fouedjio +1
  • Conference Article
  • Citations7

STAR: Noisy Semi-Supervised Transfer Learning for Visual Classification

  • Oct 20, 2021
  • Hasib Zunair +3
  • Dissertation

Active and Multi-View Machine Learning for microRNA Prediction

  • Jan 01, 2018
  • Mohsen Sheikh Hassani
  • Research Article
  • Citations27

Semisupervised Negative Correlation Learning.

  • Mar 01, 2018
  • IEEE Transactions on Neural Networks and Learning Systems
  • Huanhuan Chen +2
  • Conference Article
  • Citations7

ONSET: Opinion and Aspect Extraction System from Unlabelled Data

  • Oct 17, 2021
  • Mohiuddin Md Abdul Qudar +2
  • Book Chapter
  • Citations629

Semi-Supervised Learning

  • Jan 01, 2005
  • Tobias Scheffer
  • Dissertation

Robust Learning with Limited Labels

  • Jan 29, 2026
  • Erik Wallin
  • Conference Article
  • Citations25

S3MKL

  • Oct 25, 2010
  • Shuhui Wang +3
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.