Research map / Adversarial robustness and security / Certified and empirical robustness
Finding Biological Plausibility for Adversarially Robust Features via Metameric Tasks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use adversarially trained ResNet-50 features instead of standard-trained features to synthesize stimuli | ResNet-50 representations · Human oddity and 2AFC matching tasks across retinal eccentricities | perceptual discriminability | Helped |
| Use adversarially trained features rather than texture-model transformations | Robust and Texform stimuli · Human oddity and 2AFC matching tasks across retinal eccentricities | perceptual discriminability trend | No clear effect |
| Use adversarially robust stimuli rather than Texform stimuli | Stimuli compared with their original images · Simulated fovea and periphery IQA | DISTS | No clear effect |
| Use adversarially robust stimuli rather than Texform stimuli | Stimuli compared with their original images · Simulated fovea and periphery IQA | MSE | Mixed |
| Use adversarially trained rather than standard-trained representations | Classification of Texform stimuli · Two-class Texform classification | hit rate | Helped |
| Use standard-trained features rather than adversarially trained features to synthesize stimuli | ResNet-50 representations · Human oddity and 2AFC matching tasks across retinal eccentricities | eccentricity-dependent change in perceptual discriminability | Hurt |
Closest papers by meaning
- Understanding Robust Learning through the Lens of Representation Similarities (NeurIPS 2022 · similarity 0.89)
- Biologically Inspired Mechanisms for Adversarial Robustness (NeurIPS 2020 · similarity 0.88)
- Training on Foveated Images Improves Robustness to Adversarial Attacks (NeurIPS 2023 · similarity 0.88)
- Towards robust vision by multi-task learning on monkey visual cortex (NeurIPS 2021 · similarity 0.88)
- Generalisation in humans and deep neural networks (NeurIPS 2018 · similarity 0.87)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.