Research map / Label-efficient and robust learning / Recognition and long-tail
Soft Calibration Objectives for Neural Networks
Archit Karandikar, Nicholas Cain, Dustin Tran, Balaji Lakshminarayanan, Jonathon Shlens, Michael Curtis Mozer, Rebecca Roelofs
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 S-AvUC secondary loss (ablation) | Focal · CIFAR-100 | ECE (%): 4.66 -> 1.57 | Improved |
| add SB-ECE secondary loss (ablation) | Focal · CIFAR-100 | ECE (%): 4.66 -> 2.30 | Improved |
| add SB-ECE secondary loss (ablation) | Focal · CIFAR-10 | ECE (%): 5.45 -> 1.19 | Improved |
| add S-AvUC secondary loss (ablation) | Focal · CIFAR-10 | ECE (%): 5.45 -> 1.58 | Improved |
| add S-AvUC secondary loss (ablation) | NLL · ImageNet | ECE (%): 3.81 -> 2.26 | Improved |
| add SB-ECE secondary loss (ablation) | NLL · ImageNet | ECE (%): 3.81 -> 3.11 | Improved |
| add MMCE secondary loss (ablation) | NLL · CIFAR-100 | ECE (%): 9.10 -> 4.77 | Improved |
| add S-AvUC to Focal costs accuracy (ablation) | Focal · CIFAR-10 | accuracy (%): 95.0 -> 94.4 | Hurt |
| add post-hoc temperature scaling (ablation) | Focal · CIFAR-100 | ECE (%): 4.66 -> 6.47 | Hurt |
| add post-hoc temperature scaling (ablation) | Focal + S-AvUC · CIFAR-100 | ECE (%): 1.57 -> 4.15 | Hurt |
| add post-hoc temperature scaling (ablation) | NLL · CIFAR-100 | ECE (%): 9.10 -> 5.36 | Improved |
| add MMCE to NLL, then temperature scaling (ablation) | NLL + MMCE · ImageNet | ECE (%): 3.12 -> 2.18 | No clear effect |
Closest papers by meaning
- Rethinking Calibration of Deep Neural Networks: Do Not Be Afraid of Overconfidence (NeurIPS 2021 · similarity 0.89)
- A Benchmark Study on Calibration (ICLR 2024 · similarity 0.88)
- Uncertainty Weighted Gradients for Model Calibration (CVPR 2025 · similarity 0.88)
- Calibration Bottleneck: Over-compressed Representations are Less Calibratable (ICML 2024 · similarity 0.87)
- On Temperature Scaling and Conformal Prediction of Deep Classifiers (ICML 2025 · similarity 0.87)
- Information-theoretic Generalization Analysis for Expected Calibration Error (NeurIPS 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.
