Research map / Label-efficient and robust learning / Recognition and long-tail
Calibrating Deep Neural Networks using Focal Loss
Jishnu Mukhoti, Viveka Kulharia, Amartya Sanyal, Stuart Golodetz, Philip Torr, Puneet Dokania
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 |
|---|---|---|---|
| Replace cross-entropy with sample-dependent focal loss | Image and text classification models · CIFAR-10, CIFAR-100, Tiny-ImageNet, 20 Newsgroups, SST Binary | ECE | Improved |
| Replace cross-entropy with focal loss | CIFAR-10 classifiers · CIFAR-10 → SVHN and CIFAR-10-C | AUROC | Improved |
| Use fixed γ=3 instead of sample-dependent γ (ablation) | ResNet-50 · CIFAR-10 | ECE (%) pre temperature scaling: 1.55 -> 1.48 | Hurt |
| Increase fixed focal-loss γ from 1 to 2 to 3 (ablation) | Focal-loss classifiers · Classification datasets | ECE | Improved |
| Set focal-loss γ too high (γ≥7) (ablation) | Focal-loss training · Classification | training success | Hurt |
| Choose γ per sample based on the correct-class probability (ablation) | Focal-loss training · Multiple classification datasets and architectures | calibration | Improved |
| Use early stopping based on validation ECE with cross-entropy (ablation) | Cross-entropy training · CIFAR-10 | test error | Hurt |
| Apply temperature scaling (ablation) | Focal-loss classifiers · Image and text classification datasets | ECE | Mixed |
Closest papers by meaning
- Dual Focal Loss for Calibration (ICML 2023 · similarity 0.92)
- Uncertainty Weighted Gradients for Model Calibration (CVPR 2025 · similarity 0.90)
- A Benchmark Study on Calibration (ICLR 2024 · similarity 0.88)
- Feature Clipping for Uncertainty Calibration (AAAI 2025 · similarity 0.87)
- MaxEnt Loss: Calibrating Graph Neural Networks under Out-of-Distribution Shift (Student Abstract) (AAAI 2024 · similarity 0.87)
- Reassessing How to Compare and Improve the Calibration of Machine Learning Models (ICLR 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.
