Research map / Architectures and efficiency / RNNs and normalization
Hyperparameter Sensitivity in Deep Outlier Detection: Analysis and a Scalable Hyper-Ensemble Solution
Xueying Ding, Lingxiao Zhao, Leman Akoglu
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 |
|---|---|---|---|
| Average scores across models with different hyperparameter configurations rather than selecting a single configuration | Deep OD models · MNIST-4, Clean | AUROC: 92.39 -> 95.94 | Improved |
| Average scores across all hyperparameter configurations rather than selecting a single model | Deep OD models · MNIST-4, Polluted | AUROC: 85.73 -> 90.46 | Improved |
| Use RobOD instead of RandNet | Outlier detection · MNIST-4 | AUROC: 85.3$\pm$3.1 -> 88.0$\pm$0.0 | Improved |
| Use RobOD instead of the strongest deep baseline | Outlier detection · MNIST-5 | AUROC: 79.3$\pm$3.8 -> 81.4$\pm$0.1 | Improved |
| Use RobOD instead of the strongest deep baseline | Outlier detection · MNIST-8 | AUROC: 85.4$\pm$2.2 -> 87.8$\pm$0.0 | Improved |
| Use RobOD instead of the strongest baseline | Outlier detection · Cardio | AUROC: 94.1$\pm$0.9 -> 93.5$\pm$0.1 | Hurt |
| Use RobOD instead of the strongest baseline | Outlier detection · Thyroid | AUROC: 97.9$\pm$0.4 -> 86.1$\pm$0.6 | Hurt |
| Use RobOD instead of the strongest baseline | Outlier detection · Lympho | AUROC: 99.5$\pm$0.3 -> 98.7$\pm$0.1 | Hurt |
| Train each member on 10% of the data rather than using full-data RobOD (ablation) | RobOD · MNIST-8 | AUROC: 87.8$\pm$0.0 -> 88.2$\pm$0.1 | Improved |
| Train each member on 10% of the data rather than using full-data RobOD (ablation) | RobOD · MNIST-5 | AUROC: 81.4$\pm$0.1 -> 78.5$\pm$0.1 | Hurt |
| Train each member on 10% of the data rather than using full-data RobOD (ablation) | RobOD · Thyroid | AUROC: 86.1$\pm$0.6 -> 88.4$\pm$0.5 | Improved |
| Train ensemble members independently rather than using shared-parameter RobOD (ablation) | RobOD · MNIST-4 | AUROC: 88.0$\pm$0.0 -> 84.5$\pm$0.0 | Hurt |
Closest papers by meaning
- DOS: Diverse Outlier Sampling for Out-of-Distribution Detection (ICLR 2024 · similarity 0.89)
- ALTBI: Constructing Improved Outlier Detection Models via Optimization of Inlier-Memorization Effect (AAAI 2025 · similarity 0.88)
- ODIM: Outlier Detection via Likelihood of Under-Fitted Generative Models (ICML 2024 · similarity 0.88)
- Expecting The Unexpected: Towards Broad Out-Of-Distribution Detection (NeurIPS 2024 · similarity 0.88)
- Out-of-Distribution Detection with Prototypical Outlier Proxy (AAAI 2025 · similarity 0.88)
- Out-Of-Distribution Detection with Diversification (Provably) (NeurIPS 2024 · similarity 0.87)
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.
