Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Learning for Single-Shot Confidence Calibration in Deep Neural Networks Through Stochastic Inferences
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 variance-weighted confidence-integrated loss instead of baseline training | ResNet-34 · Tiny ImageNet val | accuracy: 50.82 -> 52.80 | Helped |
| use variance-weighted confidence-integrated loss instead of baseline training | ResNet-34 · Tiny ImageNet val | ECE: 0.067 -> 0.027 | Helped |
| use variance-weighted confidence-integrated loss instead of baseline training | VGG-16 · Tiny ImageNet val | ECE: 0.346 -> 0.053 | Helped |
| use variance-weighted confidence-integrated loss instead of baseline training | DenseNet-40-12 · Tiny ImageNet val | ECE: 0.020 -> 0.025 | Hurt |
| use variance-weighted confidence-integrated loss instead of baseline training | ResNet-34 · CIFAR-100 test | accuracy: 77.19 -> 78.64 | Helped |
| use variance-weighted confidence-integrated loss instead of baseline training | ResNet-34 · CIFAR-100 test | ECE: 0.109 -> 0.034 | Helped |
| replace variance-weighted loss with CI using a global coefficient (ablation) | VWCI classifier · Tiny ImageNet val | accuracy: 52.80 -> 51.45 | Hurt |
| replace variance-weighted loss with CI using a global coefficient (ablation) | VWCI classifier · CIFAR-100 test | ECE: 0.034 vs 0.029 | Mixed |
| initialize VWCI fine-tuning from a pretrained baseline network (ablation) | VWCI ResNet-34 · CIFAR-100 test | accuracy: 78.64 -> 77.87 | Hurt |
| initialize VWCI fine-tuning from a pretrained baseline network (ablation) | VWCI ResNet-34 · CIFAR-100 test | ECE: 0.034 -> 0.026 | Helped |
| initialize VWCI fine-tuning from a pretrained baseline network (ablation) | VWCI VGG-16 · Tiny ImageNet val | accuracy: 48.03 -> 74.17 | Helped |
| increase the number of stochastic inference samples during training (ablation) | VWCI · Tiny ImageNet and CIFAR-100 | ECE | Mixed |
Closest papers by meaning
- Improving model calibration with accuracy versus uncertainty optimization (NeurIPS 2020 · similarity 0.89)
- Uncertainty Weighted Gradients for Model Calibration (CVPR 2025 · similarity 0.87)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · similarity 0.87)
- Pointwise Information Measures as Confidence Estimators in Deep Neural Networks: A Comparative Study (ICML 2025 · similarity 0.86)
- A Benchmark Study on Calibration (ICLR 2024 · similarity 0.86)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 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.