Research map / Label-efficient and robust learning / OOD and anomaly detection
Out-of-Distribution Detection with An Adaptive Likelihood Ratio on Informative Hierarchical VAE
Yewen Li, Chaojie Wang, Xiaobo Xia, Tongliang Liu, Xin Miao, Bo An
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 |
|---|---|---|---|
| Use LLR>2 instead of LLR>1 as score (ablation) | 3-layer informative HVAE · CIFAR10/SVHN | AUROC: 68.4 -> 93.0 | Improved |
| Use LLR>2 instead of LLR>1 as score (ablation) | 3-layer informative HVAE · FashionMNIST/MNIST | AUROC: 97.5 -> 97.4 | No clear effect |
| Use LLR>k score instead of plain likelihood L (ablation) | 3-layer informative HVAE · CIFAR10/SVHN | AUROC: 49.9 -> 93.0 | Improved |
| Use LLR>k score instead of plain likelihood L (ablation) | 3-layer informative HVAE · FashionMNIST/MNIST | AUROC: 55.3 -> 97.4 | Improved |
| Use adaptive LLRada instead of fixed-k LLR>1 (ablation) | 3-layer informative HVAE · CIFAR10/SVHN | AUROC: 68.4 -> 92.6 | Improved |
| Use adaptive LLRada instead of best fixed k (LLR>2) (ablation) | 3-layer informative HVAE · CIFAR10/SVHN | AUROC: 93.0 -> 92.6 | No clear effect |
| Use informative hierarchical VAE instead of HVAE | HVAE with LLRada · FashionMNIST, KMNIST OOD | AUROC: 86.4 -> 95.0 | Improved |
| Use informative hierarchical VAE instead of BIVA | BIVA with LLRada · FashionMNIST, notMNIST OOD | AUROC: 96.4 -> 99.7 | Improved |
| Use informative hierarchical VAE instead of HVAE | HVAE with LLRada · CIFAR10, CelebA OOD | AUROC: 39.8 -> 72.1 | Improved |
| Use informative hierarchical VAE instead of HVAE | HVAE with LLRada · CIFAR10, LFWPeople OOD | AUROC: 42.5 -> 83.0 | Improved |
| Use informative hierarchical VAE instead of BIVA | BIVA with LLRada · CIFAR10, Flower102 OOD | AUROC: 61.9 -> 63.4 | No clear effect |
| Use informative hierarchical VAE instead of HVAE on already-saturated OOD sets | HVAE with LLRada · FashionMNIST, SmallNORB OOD | AUROC: 100 -> 100 | No clear effect |
Closest papers by meaning
- NECO: NEural Collapse Based Out-of-distribution detection (ICLR 2024 · similarity 0.88)
- Detecting Out-of-Distribution Through the Lens of Neural Collapse (CVPR 2025 · similarity 0.88)
- The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection (NeurIPS 2024 · similarity 0.88)
- Diffusion-based Layer-wise Semantic Reconstruction for Unsupervised Out-of-Distribution Detection (NeurIPS 2024 · similarity 0.88)
- Learning to Shape In-distribution Feature Space for Out-of-distribution Detection (NeurIPS 2024 · similarity 0.88)
- Double Descent Meets Out-of-Distribution Detection: Theoretical Insights and Empirical Analysis on the Role of Model Complexity (NeurIPS 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.
