mlsanityData to 14 Oct 2026

Inverting Adversarially Robust Networks for Image Synthesis

arXiv 2106.06927 ↗First posted 13 Jun 2021cs.CV cs.LG cs.NEcited by 6

Not on the map. Why?

Abstract (arXiv)

Despite unconditional feature inversion being the foundation of many image synthesis applications, training an inverter demands a high computational budget, large decoding capacity and imposing conditions such as autoregressive priors. To address these limitations, we propose the use of adversarially robust representations as a perceptual primitive for feature inversion. We train an adversarially robust encoder to extract disentangled and perceptually-aligned image representations, making them easily invertible. By training a simple generator with the mirror architecture of the encoder, we achieve superior reconstruction quality and generalization over standard models. Based on this, we propose an adversarially robust autoencoder and demonstrate its improved performance on style transfer, image denoising and anomaly detection tasks. Compared to recent ImageNet feature inversion methods, our model attains improved performance with significantly less complexity.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 54 of its 89 references that are in the map. First posted on arXiv 13 Jun 2021.

Builds on

54 of 89 references.

References · 54

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  4. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  5. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  6. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  7. 2017Neural Discrete Representation Learningcited by 4,712
  8. 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,527
  9. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,253
  10. 2016Perceptual Losses for Real-Time Style Transfer and Super-Resolutioncited by 2,694
  11. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,371
  12. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
38 more
  1. 2018Spectral Normalization for Generative Adversarial Networkscited by 1,636
  2. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,451
  3. 2019Generating Diverse High-Fidelity Images with VQ-VAE-2cited by 1,387
  4. 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
  5. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,268
  6. 2017Deep Image Priorcited by 1,098
  7. 2020RobustBench: a standardized adversarial robustness benchmarkcited by 661
  8. 2017FFDNet: Toward a Fast and Flexible Solution for CNN based Image Denoisingcited by 510
  9. 2014Understanding Deep Image Representations by Inverting Themcited by 403
  10. 2017Universal Style Transfer via Feature Transformscited by 354
  11. 2015Understanding Neural Networks Through Deep Visualizationcited by 334
  12. 2020Do Adversarially Robust ImageNet Models Transfer Better?cited by 316
  13. 2019Deep Semi-Supervised Anomaly Detectioncited by 291
  14. 2014Caffe: Convolutional Architecture for Fast Feature Embeddingcited by 288
  15. 2018Toward Convolutional Blind Denoising of Real Photographscited by 274
  16. 2018Deep Anomaly Detection Using Geometric Transformationscited by 269
  17. 2019The Secret Revealer: Generative Model-Inversion Attacks Against Deep Neural Networkscited by 246
  18. 2019Large Scale Adversarial Representation Learningcited by 239
  19. 2019Learning to Discover Novel Visual Categories via Deep Transfer Clusteringcited by 225
  20. 2016Generating Images with Perceptual Similarity Metrics based on Deep Networkscited by 222
  21. 2020Geometry-aware Instance-reweighted Adversarial Trainingcited by 198
  22. 2015Trainable Nonlinear Reaction Diffusion: A Flexible Framework for Fast and Effective Image Restorationcited by 195
  23. 2016Synthesizing the preferred inputs for neurons in neural networks via deep generator networkscited by 181
  24. 2019Real Image Denoising with Feature Attentioncited by 176
  25. 2015Inverting Visual Representations with Convolutional Networkscited by 170
  26. 2019Photorealistic Style Transfer via Wavelet Transformscited by 166
  27. 2018High Quality Monocular Depth Estimation via Transfer Learningcited by 133
  28. 2019Scale-Equivariant Steerable Networkscited by 110
  29. 2015Optimal whitening and decorrelationcited by 97
  30. 2015Visualizing Deep Convolutional Neural Networks Using Natural Pre-Imagescited by 93
  31. 2016High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesiscited by 90
  32. 2016Deep Feature Interpolation for Image Content Changescited by 80
  33. 2019Interpreting Adversarially Trained Convolutional Neural Networkscited by 75
  34. 2018Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Networkcited by 69
  35. 2019Understanding Neural Networks via Feature Visualization: A surveycited by 65
  36. 2019Are Perceptually-Aligned Gradients a General Property of Robust Classifiers?cited by 43
  37. 2020Semantic Pyramid for Image Generationcited by 34
  38. 2019Blind Universal Bayesian Image Denoising with Gaussian Noise Level Learningcited by 31

50 most cited of 54 All 54 (JSON)

Built on by

Cited by year

Citing papers per year: 2022: 1; 2023: 2; 2024: 1; 2025: 2.

6 papers.

  1. 2023Making Vision Transformers Truly Shift-Equivariantcited by 16
  2. 2024Shedding More Light on Robust Classifiers under the lens of Energy-based Modelscited by 9
  3. 2022MAGIC: Mask-Guided Image Synthesis by Inverting a Quasi-Robust Classifiercited by 6
  4. 2023Exploring the Connection between Robust and Generative Modelscited by 2
  5. 2025Understanding Adversarial Training with Energy-based Modelscited by 2
  6. 2025What Your Features Reveal: Data-Efficient Black-Box Feature Inversion Attack for Split DNNscited by 2