Research map / Label-efficient and robust learning / OOD and anomaly detection
Energy-based Out-of-distribution Detection
Read the paperCheck 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 |
|---|---|---|---|
| replace softmax confidence with energy score (no retraining) | pre-trained WideResNet · CIFAR-10 OOD | FPR95: 51.04 -> 33.01 | Helped |
| replace softmax confidence with energy score | pre-trained WideResNet · CIFAR-100 OOD | FPR95: 80.41 -> 73.60 | Helped |
| replace softmax with energy score | pre-trained WideResNet · CIFAR-10 vs Texture | AUROC: 88.50 -> 85.22 | Hurt |
| energy-bounded fine-tuning instead of Outlier Exposure | WideResNet · CIFAR-10 OOD | FPR95: 8.53 -> 3.32 | Helped |
| energy-bounded fine-tuning instead of Outlier Exposure | WideResNet · CIFAR-100 OOD | FPR95: 58.10 -> 47.55 | Helped |
| energy fine-tuning | WideResNet · CIFAR-100 | in-dist test error: 24.04 -> 24.58 | No clear effect |
| increase temperature T above 1 for energy score (ablation) | Energy score · CIFAR-10 OOD | FPR95 | Hurt |
| use too small in-distribution margin m_in (ablation) | Energy fine-tuning · CIFAR-10 OOD | FPR95 | Hurt |
| vary out-distribution margin m_out (ablation) | Energy fine-tuning · CIFAR-10 OOD | FPR95 | No clear effect |
Closest papers by meaning
- Dual Energy-Based Model with Open-World Uncertainty Estimation for Out-of-distribution Detection (CVPR 2025 · similarity 0.90)
- Energy-based Hopfield Boosting for Out-of-Distribution Detection (NeurIPS 2024 · similarity 0.89)
- Bounded and Uniform Energy-based Out-of-distribution Detection for Graphs (ICML 2024 · similarity 0.88)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 2025 · similarity 0.88)
- ConjNorm: Tractable Density Estimation for Out-of-Distribution Detection (ICLR 2024 · similarity 0.88)
- EDGE: Unknown-aware Multi-label Learning by Energy Distribution Gap Expansion (AAAI 2025 · similarity 0.88)
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.