Research map / Label-efficient and robust learning / OOD and anomaly detection
Secure Out-of-Distribution Task Generalization with Energy-Based Models
Shengzhuang Chen, Long-Kai Huang, Jonathan Richard Schwarz, Yilun Du, Ying Wei
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 prior EBM energy to data-EBM SNLL (energy sum) (ablation) | EBML-CNPs · Sinusoids OOD detection | FPR95: 29.40 -> 14.20 | Improved |
| Replace EBM prior with Gaussian prior in energy sum (ablation) | ABML · Sinusoids OOD detection | FPR95: 61.00 -> 52.20 | Improved |
| Add Gaussian prior NLL to SNLL (ablation) | CNPs · Sinusoids OOD detection | FPR95: 38.40 -> 34.40 | Improved |
| Add Gaussian prior NLL to SNLL (ablation) | CNPs · Sinusoids OOD detection | AUROC: 94.81 -> 94.61 | No clear effect |
| Use EBML Energy Sum over best OOD baseline | EBML · Meta-dataset Simple-CNAPs | FPR95: 43.57 -> 6.74 | Improved |
| Use EBML Energy Sum over best OOD baseline | EBML · Meta-dataset TSA | FPR95: 23.83 -> 3.64 | Improved |
| Use EBML energy sum instead of Metafun SNLL | EBML-CNPs · DrugOOD | FPR95: 65.17 -> 1.40 | Improved |
| Model joint p(X,Y) instead of conditional p(Y|X) (ablation) | EBML · DrugOOD OOD detection | detection | Improved |
| Use EBM decoder instead of Gaussian decoder (ablation) | EBML · multi-sinusoids | predictive distribution | Improved |
| Use EBML-CNPs instead of CNPs for regression | EBML-CNPs · Sinusoids | ID MSE: 0.009±0.002 | No clear effect |
| Use EBML-CNPs instead of CNPs | EBML-CNPs · multi-sinusoids | ID NLL: 0.58 lower | Improved |
| Use EBML-CNPs instead of CNPs | EBML-CNPs · DrugOOD | OOD Mean R2: 0.066 -> 0.071 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Metafun — MetaFun: Meta-Learning with Iterative Functional Updates (ICML 2020) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Meta-Learning with Less Forgetting on Large-Scale Non-stationary Task Distributions (ECCV 2022 · similarity 0.88)
- OOD-MAML: Meta-Learning for Few-Shot Out-of-Distribution Detection and Classification (NeurIPS 2020 · similarity 0.87)
- Meta Learning Low Rank Covariance Factors for Energy Based Deterministic Uncertainty (ICLR 2022 · similarity 0.87)
- Task-Robust Model-Agnostic Meta-Learning (NeurIPS 2020 · similarity 0.87)
- MetaOOD: Automatic Selection of OOD Detection Models (ICLR 2025 · similarity 0.86)
- Task-Aware Virtual Training: Enhancing Generalization in Meta-Reinforcement Learning for Out-of-Distribution Tasks (ICML 2025 · 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.
