Research map / Label-efficient and robust learning / Recognition and long-tail
DOCTOR: A Simple Method for Detecting Misclassification Errors
Federica Granese, Marco Romanelli, Daniele Gorla, Catuscia Palamidessi, Pablo Piantanida
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 |
|---|---|---|---|
| use all softmax probabilities (sum of squares) instead of only the maximum (SR) | DOCTOR D_alpha (TBB) · AmazonFashion | FRR at 95% TRR (%): 50.1 -> 27.1 | Improved |
| use all softmax probabilities instead of only the maximum (SR) | DOCTOR D_alpha (TBB) · CIFAR10 | AUROC (%): 93.8 -> 94 | Improved |
| use all softmax probabilities instead of only the maximum (SR) | DOCTOR D_alpha (TBB) · TinyImageNet | AUROC (%): 84.9 -> 84.9 | No clear effect |
| add input perturbation (PBB) to DOCTOR D_alpha (ablation) | DOCTOR D_alpha · CIFAR10 | FRR at 95% TRR (%): 17.9 -> 13.9 | Improved |
| add input perturbation (PBB) to DOCTOR D_alpha (ablation) | DOCTOR D_alpha · CIFAR10 | AUROC (%): 94 -> 95.2 | Improved |
| use D_alpha instead of D_beta in TBB (ablation) | DOCTOR · CIFAR10 | AUROC (%): 68.5 -> 94 | Improved |
| use D_alpha instead of D_beta in PBB (ablation) | DOCTOR · CIFAR10 | AUROC (%): 94.8 -> 95.2 | Improved |
| use DOCTOR instead of ODIN in PBB | DOCTOR D_alpha · CIFAR10 | FRR at 95% TRR (%): 18.4 -> 13.9 | Improved |
| use DOCTOR instead of ODIN in PBB | DOCTOR D_alpha · CIFAR100 | AUROC (%): 87.1 -> 88.2 | Improved |
| use DOCTOR instead of Mahalanobis (MHLNB) in PBB | DOCTOR D_alpha · CIFAR100 | AUROC (%): 50 -> 88.2 | Improved |
| use DOCTOR instead of ODIN with OOD samples mixed in | DOCTOR D_alpha · CIFAR10 + iSUN (1 in 5 OOD) | AUROC (%): 94.6 -> 95.4 | Improved |
Closest papers by meaning
- Model Doctor: A Simple Gradient Aggregation Strategy for Diagnosing and Treating CNN Classifiers (AAAI 2022 · similarity 0.87)
- Interpretable Explanations of Black Boxes by Meaningful Perturbation (ICCV 2017 · similarity 0.84)
- Interpreting Image Classifiers by Generating Discrete Masks (TPAMI 2020 · similarity 0.84)
- A Decade's Battle on Dataset Bias: Are We There Yet? (ICLR 2025 · similarity 0.83)
- Towards Trustworthy Deep Learning (AAAI 2024 · similarity 0.82)
- Mitigating Biases in Blackbox Feature Extractors for Image Classification Tasks (NeurIPS 2024 · similarity 0.82)
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.
