Research map / Image generation / Autoregressive and latent generation
Generalisation in humans and deep neural networks
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 |
|---|---|---|---|
| train directly on a specific distortion | ResNet-50 · 16-class-ImageNet, same distortion at test | accuracy | Helped |
| train on one distortion, test on a different distortion | ResNet-50 · 16-class-ImageNet | accuracy | Mixed |
| train on salt-and-pepper noise, test on uniform noise (or vice versa) | ResNet-50 · 16-class-ImageNet | accuracy | No clear effect |
| add uniform noise to training data alongside another distortion | ResNet-50 · 16-class-ImageNet | accuracy | No clear effect |
| train on all distortions except the test noise | ResNet-50 · 16-class-ImageNet, held-out noise | accuracy: 11 to 14% | No clear effect |
| train longer (200 vs 100 epochs) | ResNet-50 on uniform noise · 16-class-ImageNet uniform noise | accuracy | Helped |
Closest papers by meaning
- Harmonizing the object recognition strategies of deep neural networks with humans (NeurIPS 2022 · similarity 0.90)
- Towards robust vision by multi-task learning on monkey visual cortex (NeurIPS 2021 · similarity 0.89)
- Minimal Images in Deep Neural Networks: Fragile Object Recognition in Natural Images (ICLR 2019 · similarity 0.89)
- Partial success in closing the gap between human and machine vision (NeurIPS 2021 · similarity 0.89)
- Interpreting Robustness Proofs of Deep Neural Networks (ICLR 2024 · similarity 0.87)
- Can Biases in ImageNet Models Explain Generalization? (CVPR 2024 · similarity 0.87)
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.