mlsanityData to 14 Oct 2026

DNNShield: Dynamic Randomized Model Sparsification, A Defense Against Adversarial Machine Learning

arXiv 2208.00498 ↗First posted 31 Jul 2022cs.CR cs.AR cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

DNNs are known to be vulnerable to so-called adversarial attacks that manipulate inputs to cause incorrect results that can be beneficial to an attacker or damaging to the victim. Recent works have proposed approximate computation as a defense mechanism against machine learning attacks. We show that these approaches, while successful for a range of inputs, are insufficient to address stronger, high-confidence adversarial attacks. To address this, we propose DNNSHIELD, a hardware-accelerated defense that adapts the strength of the response to the confidence of the adversarial input. Our approach relies on dynamic and random sparsification of the DNN model to achieve inference approximation efficiently and with fine-grain control over the approximation error. DNNSHIELD uses the output distribution characteristics of sparsified inference compared to a dense reference to detect adversarial inputs. We show an adversarial detection rate of 86% when applied to VGG16 and 88% when applied to ResNet50, which exceeds the detection rate of the state of the art approaches, with a much lower overhead. We demonstrate a software/hardware-accelerated FPGA prototype, which reduces the performance impact of DNNSHIELD relative to software-only CPU and GPU implementations.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 44 of its 59 references that are in the map. First posted on arXiv 31 Jul 2022.

Builds on

44 of 59 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  5. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  6. 2013Intriguing properties of neural networkscited by 5,155
  7. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  8. 2016Adversarial examples in the physical worldcited by 1,769
  9. 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
  10. 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,318
  11. 2015The Limitations of Deep Learning in Adversarial Settingscited by 790
  12. 2016Universal adversarial perturbationscited by 784
32 more
  1. 2015Distillation as a Defense to Adversarial Perturbations against Deep Neural Networkscited by 678
  2. 2020On Adaptive Attacks to Adversarial Example Defensescited by 555
  3. 2017Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methodscited by 546
  4. 2018Certified Robustness to Adversarial Examples with Differential Privacycited by 468
  5. 2018Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Surveycited by 460
  6. 2019On Evaluating Adversarial Robustnesscited by 447
  7. 2017Mitigating Adversarial Effects Through Randomizationcited by 383
  8. 2018Characterizing Adversarial Subspaces Using Local Intrinsic Dimensionalitycited by 282
  9. 2017Detecting Adversarial Samples from Artifactscited by 245
  10. 2018Stochastic Activation Pruning for Robust Adversarial Defensecited by 171
  11. 2017EAD: Elastic-Net Attacks to Deep Neural Networks via Adversarial Examplescited by 155
  12. 2017SCNN: An Accelerator for Compressed-sparse Convolutional Neural Networkscited by 135
  13. 2018Parametric Noise Injection: Trainable Randomness to Improve Deep Neural Network Robustness against Adversarial Attackcited by 125
  14. 2019The Odds are Odd: A Statistical Test for Detecting Adversarial Examplescited by 75
  15. 2017Mitigating Evasion Attacks to Deep Neural Networks via Region-based Classificationcited by 56
  16. 2017Detecting Adversarial Image Examples in Deep Networks with Adaptive Noise Reductioncited by 49
  17. 2019A New Defense Against Adversarial Images: Turning a Weakness into a Strengthcited by 49
  18. 2019ALWANN: Automatic Layer-Wise Approximation of Deep Neural Network Accelerators without Retrainingcited by 28
  19. 2020Learn2Perturb: an End-to-end Feature Perturbation Learning to Improve Adversarial Robustnesscited by 27
  20. 2019Adversarial Defense Through Network Profiling Based Path Extractioncited by 24
  21. 2018Accelerator-Aware Pruning for Convolutional Neural Networkscited by 22
  22. 2017Attacking the Madry Defense Model with L1-based Adversarial Examplescited by 21
  23. 2020TFApprox: Towards a Fast Emulation of DNN Approximate Hardware Accelerators on GPUcited by 17
  24. 2020TensorDash: Exploiting Sparsity to Accelerate Deep Neural Network Training and Inferencecited by 16
  25. 2020QUANOS- Adversarial Noise Sensitivity Driven Hybrid Quantization of Neural Networkscited by 12
  26. 2020Defensive Approximation: Securing CNNs using Approximate Computingcited by 12
  27. 2020Ptolemy: Architecture Support for Robust Deep Learningcited by 12
  28. 2018Bypassing Feature Squeezing by Increasing Adversary Strengthcited by 8
  29. 2019Are Odds Really Odd? Bypassing Statistical Detection of Adversarial Examplescited by 7
  30. 20212-in-1 Accelerator: Enabling Random Precision Switch for Winning Both Adversarial Robustness and Efficiencycited by 5
  31. 2018On the Limitation of MagNet Defense against L1-based Adversarial Examplescited by 3
  32. 2021RingCNN: Exploiting Algebraically-Sparse Ring Tensors for Energy-Efficient CNN-Based Computational Imagingcited by 3

Built on by

Not cited yet.