Research map / Image generation / GANs
Difficulty-Aware Simulator for Open Set Recognition
WonJun Moon, Junho Park, Hyun Seok Seong, Cheol-Ho Cho, Jae-Pil Heo
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 |
|---|---|---|---|
| add Copycat hard fakes (ablation) | classifier without fakes · CIFAR100 openness 22.54% | AUROC (%): 72.28 -> 76.59 | Improved |
| add GAN moderate fakes (ablation) | classifier without fakes · CIFAR100 openness 68.38% | AUROC (%): 77.34 -> 81.13 | Improved |
| add Copycat easy fakes (ablation) | classifier without fakes · CIFAR100 openness 22.54% | AUROC (%): 72.28 -> 72.45 | No clear effect |
| add GAN and easy fakes to Copycat hard fakes (ablation) | DIAS · CIFAR100 openness 68.38% | AUROC (%): 80.91 -> 83.95 | Improved |
| add GAN fakes to classifier (F1) (ablation) | classifier without fakes · CIFAR10 | macro F1: 0.704 -> 0.777 | Improved |
| add Copycat to classifier (F1) (ablation) | classifier without fakes · CIFAR+50 | macro F1: 0.660 -> 0.816 | Improved |
| add Copycat to GAN fakes (F1) (ablation) | DIAS · CIFAR+50 | macro F1: 0.775 -> 0.829 | Improved |
| use DIAS instead of Softmax | classifier · Tiny-ImageNet closed-set accuracy | accuracy: 0.653 -> 0.700 | Improved |
| use DIAS instead of ARPL+cs | DIAS · CIFAR10 | AUROC: 0.819 -> 0.850 | Improved |
| use DIAS instead of CVAECap | DIAS · SVHN | AUROC: 0.956 -> 0.943 | Hurt |
| use inherent threshold instead of searched best threshold (ablation) | DIAS · MNIST | macro F1: 0.970 -> 0.953 | Hurt |
| set alpha_hard to 1 or 0 instead of 0.5 (ablation) | DIAS · Open set (alpha_hard sweep) | AUROC | Hurt |
Closest papers by meaning
- OpenGAN: Open-Set Recognition via Open Data Generation (ICCV 2021 · similarity 0.85)
- Open Set Learning with Counterfactual Images (ECCV 2018 · similarity 0.85)
- Open-Set Recognition: A Good Closed-Set Classifier is All You Need (ICLR 2022 · similarity 0.84)
- The Devil is in the Wrongly-classified Samples: Towards Unified Open-set Recognition (ICLR 2023 · similarity 0.84)
- From Coarse to Fine-Grained Open-Set Recognition (CVPR 2024 · similarity 0.84)
- Unlocking Better Closed-Set Alignment Based on Neural Collapse for Open-Set Recognition (AAAI 2025 · similarity 0.83)
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.
