Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Contextual Similarity Distillation: Ensemble Uncertainties with a Single Model
Moritz Akiya Zanger, Pascal R. Van der Vaart, Wendelin Boehmer, Matthijs T. J. Spaan
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 data augmentations as context data (ablation) | CSD · MNIST-family distribution shift | AUROC: 96.63 -> 98.22 | Improved |
| add unlabeled target-distribution context data (ablation) | CSD · MNIST-family distribution shift | AUROC: 96.63 -> 98.57 | Improved |
| add target-distribution context data on top of augmentations (ablation) | CSD-Aug. · MNIST-family distribution shift | AUROC: 98.22 -> 98.57 | Improved |
| add augmentations to context data (ablation) | CSD · MNIST-family distribution shift | AUPR-OUT: 96.19 -> 97.80 | Improved |
| add target-distribution context data (ablation) | CSD · MNIST-family distribution shift | AUPR-IN: 96.94 -> 98.86 | Improved |
| add augmentations to context data (ablation) | CSD · MNIST-family distribution shift | test accuracy: 96.29 -> 96.28 | No clear effect |
| single-model CSD instead of 15-member deep ensemble | ENS(15) · MNIST-family distribution shift | AUROC: 94.00 -> 96.63 | Improved |
| single-model CSD instead of RND | RND · MNIST-family distribution shift | AUROC: 94.40 -> 96.63 | Improved |
| single-model CSD instead of ensemble of 15 on accuracy | ENS(15) · MNIST-family distribution shift | test accuracy: 97.18 -> 96.29 | Hurt |
| increase ensemble size from 3 to 15 (ablation) | deep ensemble · MNIST-family distribution shift | AUROC: 92.30 -> 94.00 | Improved |
Closest papers by meaning
- Knowledge Distillation of Uncertainty using Deep Latent Factor Model (NeurIPS 2025 · similarity 0.87)
- DomED: Redesigning Ensemble Distillation for Domain Generalization (ICML 2026 · similarity 0.86)
- Discriminative Jackknife: Quantifying Uncertainty in Deep Learning via Higher-Order Influence Functions (ICML 2020 · similarity 0.86)
- Neural Network Approximation for Pessimistic Offline Reinforcement Learning (AAAI 2024 · similarity 0.85)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.85)
- Understanding and Diagnosing Deep Reinforcement Learning (ICML 2024 · 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.
