mlsanityData to 14 Oct 2026

Ensemble Methods as a Defense to Adversarial Perturbations Against Deep Neural Networks

arXiv 1709.03423 ↗First posted 11 Sep 2017stat.ML cs.LGcited by 31

Not on the map. Why?

Abstract (arXiv)

Deep learning has become the state of the art approach in many machine learning problems such as classification. It has recently been shown that deep learning is highly vulnerable to adversarial perturbations. Taking the camera systems of self-driving cars as an example, small adversarial perturbations can cause the system to make errors in important tasks, such as classifying traffic signs or detecting pedestrians. Hence, in order to use deep learning without safety concerns a proper defense strategy is required. We propose to use ensemble methods as a defense strategy against adversarial perturbations. We find that an attack leading one model to misclassify does not imply the same for other networks performing the same task. This makes ensemble methods an attractive defense strategy against adversarial attacks. We empirically show for the MNIST and the CIFAR-10 data sets that ensemble methods not only improve the accuracy of neural networks on test data but also increase their robustness against adversarial perturbations.

Key facts. Cited by 31 papers in the map (data through 14 Oct 2026). First posted on arXiv 11 Sep 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 12; 2022: 5; 2023: 6; 2024: 2; 2025: 4; 2026: 2.

31 papers.

Sort:

By area:

  1. 2021Fixing Data Augmentation to Improve Adversarial Robustnesscited by 204
  2. 2021Data Augmentation Can Improve Robustnesscited by 194
  3. 2021Relating Adversarially Robust Generalization to Flat Minimacited by 57
  4. 2021Deep Ensembling with No Overhead for either Training or Testing: The All-Round Blessings of Dynamic Sparsitycited by 44
  5. 2023Towards Trustworthy and Aligned Machine Learning: A Data-centric Survey with Causality Perspectivescited by 30
  6. 2021Knowledge Enhanced Machine Learning Pipeline against Diverse Adversarial Attackscited by 24
  7. 2021Security and Privacy for Artificial Intelligence: Opportunities and Challengescited by 16
  8. 2024Spiking Convolutional Neural Networks for Text Classificationcited by 10
  9. 2021Neural Ensemble Search via Bayesian Samplingcited by 7
  10. 2023To Stay or Not to Stay in the Pre-train Basin: Insights on Ensembling in Transfer Learningcited by 6
  11. 20212-in-1 Accelerator: Enabling Random Precision Switch for Winning Both Adversarial Robustness and Efficiencycited by 5
  12. 2022Sequential Bayesian Neural Subnetwork Ensemblescited by 5
19 more
  1. 2024Towards Adversarial Robustness of Model-Level Mixture-of-Experts Architectures for Semantic Segmentationcited by 5
  2. 2022Resisting Adversarial Attacks in Deep Neural Networks using Diverse Decision Boundariescited by 4
  3. 2021Tensor Normalization and Full Distribution Trainingcited by 3
  4. 2023Alignment is not sufficient to prevent large language models from generating harmful information: A psychoanalytic perspectivecited by 3
  5. 2025Advancing Robustness in Deep Reinforcement Learning with an Ensemble Defense Approachcited by 3
  6. 2021Ensemble Defense with Data Diversity: Weak Correlation Implies Strong Robustnesscited by 2
  7. 2021Saliency Diversified Deep Ensemble for Robustness to Adversariescited by 2
  8. 2022Memory Defense: More Robust Classification via a Memory-Masking Autoencodercited by 2
  9. 2025Robust Confinement State Classification with Uncertainty Quantification through Ensembled Data-Driven Methodscited by 2
  10. 2021Resilience from Diversity: Population-based approach to harden models against adversarial attackscited by 1
  11. 2025Divided We Fall: Defending Against Adversarial Attacks via Soft-Gated Fractional Mixture-of-Experts with Randomized Adversarial Trainingcited by 1
  12. 2022Defending Adversarial Examples by Negative Correlation Ensemblecited by 0
  13. 2022Butterfly Effect Attack: Tiny and Seemingly Unrelated Perturbations for Object Detectioncited by 0
  14. 2023Reducing Adversarial Training Cost with Gradient Approximationcited by 0
  15. 2023Improving the Robustness of Quantized Deep Neural Networks to White-Box Attacks using Stochastic Quantization and Information-Theoretic Ensemble Trainingcited by 0
  16. 2023May the Noise be with you: Adversarial Training without Adversarial Examplescited by 0
  17. 2025Lightweight Defense Against Adversarial Attacks in Time Series Classificationcited by 0
  18. 2026QShield: Securing Neural Networks Against Adversarial Attacks using Quantum Circuitscited by 0
  19. 2026Towards One-for-All Robustness Across a Continuum of Threat Levelscited by 0