Research map / Label-efficient and robust learning / OOD and anomaly detection
Hybrid Energy Based Model in the Feature Space for Out-of-Distribution Detection
Marc Lafon, Elias Ramzi, Clément Rambour, Nicolas THOME
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 a learned residual EBM to GMM (ablation) | GMM prior scorer · CIFAR-10, average OOD sets | FPR95: 35.1 -> 31.5 | Improved |
| Add a learned residual EBM to GMM (ablation) | GMM prior scorer · CIFAR-100, average OOD sets | FPR95: 66.2 -> 64.9 | Improved |
| Add a learned residual EBM to EL (ablation) | EL prior scorer · CIFAR-10, average OOD sets | FPR95: 32.6 -> 31.0 | Improved |
| Add a learned residual EBM to EL (ablation) | EL prior scorer · CIFAR-100, average OOD sets | FPR95: 74.3 -> 72.6 | Improved |
| Train a residual EBM over a GMM prior rather than use a fully data-driven EBM (ablation) | OOD density estimation · CIFAR-10, average OOD sets | AUC: 93.3 -> 93.8 | Improved |
| Train a residual EBM over a GMM prior rather than use a fully data-driven EBM (ablation) | OOD density estimation · CIFAR-100, average OOD sets | AUC: 79.5 -> 82.1 | Improved |
| Compose corrected GMM and std-pooled GMM energies (ablation) | HEAT · CIFAR-10, average OOD sets | FPR95: 31.5 -> 25.6 | Improved |
| Compose corrected GMM and std-pooled GMM energies (ablation) | HEAT · CIFAR-100, average OOD sets | AUC: 82.1 -> 82.7 | Improved |
| Compose corrected GMM, std-pooled GMM, and EL energies (ablation) | HEAT · CIFAR-10, average OOD sets | FPR95: 31.5 -> 23.5 | Improved |
| Compose corrected GMM, std-pooled GMM, and EL energies (ablation) | HEAT · CIFAR-100, average OOD sets | AUC: 82.1 -> 83.0 | Improved |
| Use HEAT instead of SSD+ | OOD detection · CIFAR-10, average OOD sets | FPR95: 35.1 -> 23.5 | Improved |
| Use HEAT instead of DICE | OOD detection · ImageNet, average OOD sets | AUC: 90.9 -> 92.6 | Improved |
Closest papers by meaning
- Energy-based Out-of-distribution Detection (NeurIPS 2020 · similarity 0.89)
- Dual Energy-Based Model with Open-World Uncertainty Estimation for Out-of-distribution Detection (CVPR 2025 · similarity 0.88)
- Enhancing the Power of OOD Detection via Sample-Aware Model Selection (CVPR 2024 · similarity 0.87)
- Energy-based Hopfield Boosting for Out-of-Distribution Detection (NeurIPS 2024 · similarity 0.86)
- ConjNorm: Tractable Density Estimation for Out-of-Distribution Detection (ICLR 2024 · similarity 0.86)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 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.
