Research map / Label-efficient and robust learning / Recognition and long-tail
OpenGAN: Open-Set Recognition via Open Data Generation
Shu Kong, Deva Ramanan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| build the discriminator on off-the-shelf features instead of pixels (ablation) | OpenGAN-0 · Setup-I SVHN | AUROC: .881 -> .988 | Improved |
| build the nearest-neighbour detector on features instead of pixels (ablation) | NN · Setup-I SVHN | AUROC: .534 -> .888 | Improved |
| train on features instead of pixels (ablation) | OpenGAN-0 · Setup-I MNIST | AUROC: .987 -> .999 | Improved |
| train on features instead of pixels (ablation) | OpenGAN-0 · Setup-I CIFAR | AUROC: .971 -> .973 | No clear effect |
| use BiGAN discriminator instead of reconstruction error (features) (ablation) | BiGAN · Setup-I SVHN | AUROC: .976 -> .993 | Improved |
| use BiGAN discriminator instead of reconstruction error (pixels) (ablation) | BiGAN · Setup-I SVHN | AUROC: .822 -> .880 | Improved |
| train BiGAN discriminator on features instead of pixels (ablation) | BiGAN · Setup-I SVHN | AUROC: .880 -> .993 | Improved |
| add GAN-generated fake open data and discriminator instead of binary classifier (features) (ablation) | OpenGAN · Setup-II cross-dataset | AUROC: .918 -> .984 | Improved |
| add GAN-generated fake open data instead of binary classifier (pixels) (ablation) | OpenGAN · Setup-II cross-dataset | AUROC: .857 -> .969 | Improved |
| train OpenGAN on pixels instead of features (ablation) | OpenGAN · Setup-II cross-dataset | AUROC: .984 -> .969 | Hurt |
| add real open training data (OpenGAN vs OpenGAN-0, features) (ablation) | OpenGAN · Setup-II cross-dataset avg | AUROC | No clear effect |
| add GAN generator and discriminator over binary classifier (features) (ablation) | OpenGAN · Setup-III Cityscapes | AUROC: .861 -> .885 | Improved |
Closest papers by meaning
- Open-Set Recognition: A Good Closed-Set Classifier is All You Need (ICLR 2022 · similarity 0.90)
- Generative-Discriminative Feature Representations for Open-Set Recognition (CVPR 2020 · similarity 0.89)
- Hybrid Models for Open Set Recognition (ECCV 2020 · similarity 0.89)
- From Coarse to Fine-Grained Open-Set Recognition (CVPR 2024 · similarity 0.87)
- All Beings Are Equal in Open Set Recognition (AAAI 2024 · similarity 0.86)
- Unlocking Better Closed-Set Alignment Based on Neural Collapse for Open-Set Recognition (AAAI 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.
