mlsanityData to 7 Oct 2026

Built-in Elastic Transformations for Improved Robustness

arXiv 2107.09391 ↗First posted 20 Jul 2021cs.CV cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We focus on building robustness in the convolutions of neural visual classifiers, especially against natural perturbations like elastic deformations, occlusions and Gaussian noise. Existing CNNs show outstanding performance on clean images, but fail to tackle naturally occurring perturbations. In this paper, we start from elastic perturbations, which approximate (local) view-point changes of the object. We present elastically-augmented convolutions (EAConv) by parameterizing filters as a combination of fixed elastically-perturbed bases functions and trainable weights for the purpose of integrating unseen viewpoints in the CNN. We show on CIFAR-10 and STL-10 datasets that the general robustness of our method on unseen occlusion, zoom, rotation, image cut and Gaussian perturbations improves, while significantly improving the performance on clean images without any data augmentation.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 22 of its 30 references that are in the map. First posted on arXiv 20 Jul 2021.

Builds on

22 of 30 references.

References · 22

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  2. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  3. 2016Wide Residual Networkscited by 3,057
  4. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  5. 2017Deformable Convolutional Networkscited by 1,481
  6. 2015Spatial Transformer Networkscited by 1,423
  7. 2018Robustness May Be at Odds with Accuracycited by 873
  8. 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 564
  9. 2018Albumentations: fast and flexible image augmentationscited by 558
  10. 2020Improving robustness against common corruptions by covariate shift adaptationcited by 427
  11. 2018Why do deep convolutional networks generalize so poorly to small image transformations?cited by 198
  12. 2018Do CIFAR-10 Classifiers Generalize to CIFAR-10?cited by 197
10 more
  1. 2019Scale-Equivariant Steerable Networkscited by 110
  2. 2017A Study and Comparison of Human and Deep Learning Recognition Performance Under Visual Distortionscited by 107
  3. 2019Interpreting Adversarially Trained Convolutional Neural Networkscited by 75
  4. 2020Revisiting Batch Normalization for Improving Corruption Robustnesscited by 55
  5. 2020Learning perturbation sets for robust machine learningcited by 48
  6. 2016Structured Receptive Fields in CNNscited by 32
  7. 2020Scale Equivariance Improves Siamese Trackingcited by 32
  8. 2020Model-Based Robust Deep Learning: Generalizing to Natural, Out-of-Distribution Datacited by 27
  9. 2021Natural Perturbed Training for General Robustness of Neural Network Classifierscited by 2
  10. 2019Adversarial Neural Pruning with Latent Vulnerability Suppressioncited by 1

Built on by

Not cited yet.