Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Energy-Based Open-World Uncertainty Modeling for Confidence Calibration
Yezhen Wang, Bo Li, Tong Che, Kaiyang Zhou, Ziwei Liu, Dongsheng Li
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 EOW-Softmax K+1-way energy-based objective to vanilla training | ResNet50 · CIFAR-100 | ECE (%): 19.1 -> 1.08 | Improved |
| add EOW-Softmax to vanilla training | VGG11 · CIFAR-10 | ECE (%): 6.30 -> 1.57 | Improved |
| add EOW-Softmax to vanilla training | ResNet50 · Tiny-ImageNet | ECE (%): 25.2 -> 3.45 | Improved |
| add EOW-Softmax to vanilla training | EOW-Softmax · CIFAR-10/CIFAR-100/Tiny-ImageNet | accuracy (%): 90.48 -> 90.24 | No clear effect |
| use EOW-Softmax instead of DBLE | VGG11 · CIFAR-10 | ECE (%): 1.50 -> 1.57 | Hurt |
| use EOW-Softmax instead of Temperature Scaling | ResNet50 · CIFAR-100 | accuracy (%): 69.84 -> 71.33 | Improved |
| add EOW-Softmax for OOD rejection by confidence threshold | ResNet50 · CIFAR-100 + CIFAR-10 OOD (threshold 0.75) | accuracy (%): 58.90 -> 69.00 | Improved |
| add EOW-Softmax for OOD rejection by confidence threshold | ResNet50 · CIFAR-100 + SVHN OOD (threshold 0.75) | accuracy (%): 26.90 -> 32.68 | Improved |
| set lambda to 0.1 instead of 1 (ablation) | EOW-Softmax · CIFAR-10 | ECE (%): 2.11 -> 1.57 | Improved |
| set lambda to 0.1 instead of 0.01 (ablation) | EOW-Softmax · CIFAR-10 | ECE (%): 3.37 -> 1.57 | Improved |
| set lambda to 0.1 instead of 1 (ablation) | EOW-Softmax · CIFAR-100 | ECE (%): 1.33 -> 1.08 | Improved |
| set lambda to 0.1 instead of 0.01 (ablation) | EOW-Softmax · CIFAR-100 | ECE (%): 2.31 -> 1.08 | Improved |
Closest papers by meaning
- Uncertainty Calibration with Energy Based Instance-wise Scaling in the Wild Dataset (ECCV 2024 · similarity 0.89)
- Density-Softmax: Efficient Test-time Model for Uncertainty Estimation and Robustness under Distribution Shifts (ICML 2024 · similarity 0.87)
- Rethinking Approximate Gaussian Inference in Classification (NeurIPS 2025 · similarity 0.85)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.85)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · similarity 0.85)
- The Minimax Rate of Second-Order Calibration (NeurIPS 2026 · similarity 0.85)
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.
