Image Synthesis with a Single (Robust) Classifier

By Shibani Santurkar, Andrew Ilyas, Dimitris Tsipras, Logan Engstrom, Brandon Tran, Aleksander Madry. 2019

217 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models (2021, neutral): “While developing the people ﬁlter, we were aiming to detect all people in all types of environments reliably, a task which is often difﬁcult for modern face detection systems especially when dealing with people of all demographics (Buolamwini & Gebru, 2018; Santurkar et al., 2019).”
Diffusion Models Beat GANs on Image Synthesis (2021, neutral): “[48] demonstrate that a robust image classifier can be used as a”
Automatic Shortcut Removal for Self-Supervised Representation Learning (2020, background): “Somewhat related, Santurkar et al. (2019) present evidence that adversarially trained classiﬁers learn more abstract semantic features.”
Understanding and Improving Fast Adversarial Training (2020, background): “[32] Shibani Santurkar, Dimitris Tsipras, Brandon Tran, Andrew Ilyas, Logan Engstrom, and Aleksander Madry.”
Hybrid Discriminative-Generative Training via Contrastive Learning (2020, background): “• Adv Training [30, 43] which proposes to use robust optimization to train classifier to be robust to the norm through which it is being attacked.”
Network-to-Network Translation with Conditional Invertible Neural Networks (2020, unknown): “[ 81, 45] ﬁnd images which maximally activate neurons of pre-trained networks and [60] shows that improved synthesis results are obtained with adversarially robust classiﬁers.”
A Survey on Audio Diffusion Models: Text To Speech Synthesis and Enhancement in Generative AI (2023, unknown): “Without requiring the pairs of clean and degraded images, unsupervised image restoration [21, 35, 77, 92] has achieved impressive results, but most of them [21, 77, 92]”
Advances in adversarial attacks and defenses in computer vision: A survey (2021, methodology): “[289] to use adversarially robust models to perform visually appealing image synthesis.”
Boosting Adversarial Training with Hypersphere Embedding (2020, background): “Besides, previous studies observe that the adversarial examples against robust models exhibit salient data characteristics [30, 61, 68, 71, 92].”
See through Gradients: Image Batch Recovery via GradInversion (2021, background): “[42] extended the approach to more realistic images.”
Making Sense of CNNs: Interpreting Deep Representations & Their Invariances with INNs (2020, background): “[51] showed that the quality of images which maximally activate certain neurons is significantly improved when activating neurons of an adversarially robust classifer.”
Intermediate Layer Optimization for Inverse Problems using Deep Generative Models (2021, methodology): “Our method uses gradients from classiﬁers trained to achieve robustness to adversarial attacks as proposed in (Santurkar et al., 2019), but guiding generative latent codes as opposed to pixels directly.”
Partial success in closing the gap between human and machine vision (2021, background): “The resulting models were found to transfer better, have meaningful gradients [65], and enable interpolating between two input images [66]: “robust optimization can actually be viewed as inducing a human prior over the features that models are able to learn” [67, p. 10].”
The Curious Case of Adversarially Robust Models: More Data Can Help, Double Descend, or Hurt Generalization (2020, background): “There is a large body of work dedicated to adversarially robust models [Zhang and Zhu, 2019, Santurkar et al., 2019, Zhang et al., 2019, Bhagoji et al., 2019, Diochnos et al., 2019, Wei and Ma, 2019].”
Denoising Diffusion Restoration Models (2022, background): “Under this unsupervised setting, priors based on deep neural networks have demonstrated impressive empirical results in various image restoration tasks [40, 50, 43, 38, 15].”