Research map / Label-efficient and robust learning / OOD and anomaly detection
Deep Hybrid Models for Out-of-Distribution Detection
Senqi Cao, Zhongfei Zhang
Check 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 |
|---|---|---|---|
| Use the full DHM rather than a vanilla DNN | Wide ResNet-28-10 · CIFAR-10 ID / CIFAR-100 OoD | AUROC: 0.868 ± 0.01 -> 1.000 ± 0.00 | Improved |
| Use the full DHM rather than a vanilla DNN | Wide ResNet-28-10 · CIFAR-100 ID / CIFAR-10 OoD | AUROC: 0.806 ± 0.01 -> 1.000 ± 0.00 | Improved |
| Use the full DHM rather than a vanilla DNN | XLNet · CLINC150 out-of-scope | AUROC: 0.975 ± 0.02 -> 1.000 ± 0.00 | Improved |
| Add a normalizing flow to the spectrally normalized classifier (ablation) | DNN+SN · CIFAR-10 ID / CIFAR-100 OoD | AUROC: 0.872 ± 0.01 -> 1.000 ± 0.00 | Improved |
| Add a normalizing flow to the spectrally normalized classifier (ablation) | DNN+SN · CIFAR-100 ID / CIFAR-10 OoD | AUROC: 0.811 ± 0.02 -> 1.000 ± 0.00 | Improved |
| Add a normalizing flow to the spectrally normalized classifier (ablation) | DNN+SN · CLINC150 out-of-scope | AUROC: 0.976 ± 0.01 -> 1.000 ± 0.00 | Improved |
| Add spectral normalization to the classifier with a normalizing flow (ablation) | DNN+NF · CIFAR-10 ID / CIFAR-100 OoD | AUROC: 0.951 ± 0.03 -> 1.000 ± 0.00 | Improved |
| Add spectral normalization to the classifier with a normalizing flow (ablation) | DNN+NF · CIFAR-100 ID / CIFAR-10 OoD | AUROC: 0.945 ± 0.04 -> 1.000 ± 0.00 | Improved |
| Add spectral normalization to the classifier with a normalizing flow (ablation) | DNN+NF · CLINC150 out-of-scope | AUROC: 0.991 ± 0.01 -> 1.000 ± 0.00 | Improved |
| Add spectral normalization to a vanilla classifier (ablation) | Wide ResNet-28-10 · CIFAR-10 ID / CIFAR-100 OoD | AUROC: 0.868 ± 0.01 -> 0.872 ± 0.01 | No clear effect |
| Add spectral normalization to a vanilla classifier (ablation) | Wide ResNet-28-10 · CIFAR-100 ID / CIFAR-10 OoD | AUROC: 0.806 ± 0.01 -> 0.811 ± 0.02 | No clear effect |
| Add a normalizing flow without spectral normalization (ablation) | vanilla classifier · CIFAR-10 ID / CIFAR-100 OoD | AUROC: 0.868 ± 0.01 -> 0.951 ± 0.03 | Improved |
Closest papers by meaning
- A Statistical Framework for Efficient Out of Distribution Detection in Deep Neural Networks (ICLR 2022 · similarity 0.91)
- Understanding Likelihood of Normalizing Flow and Image Complexity through the Lens of Out-of-Distribution Detection (AAAI 2024 · similarity 0.89)
- Splitting & Integrating: Out-of-Distribution Detection via Adversarial Gradient Attribution (ICML 2025 · similarity 0.89)
- ProHOC: Probabilistic Hierarchical Out-of-Distribution Classification via Multi-Depth Networks (CVPR 2025 · similarity 0.88)
- Expecting The Unexpected: Towards Broad Out-Of-Distribution Detection (NeurIPS 2024 · similarity 0.88)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 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.
