Research map / Probabilistic and scientific modeling / Variational and generative sampling
Further Analysis of Outlier Detection with Deep Generative Models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 white-noise Box-Pierce test instead of a single-sided likelihood test | AR-DGM outlier detection · CIFAR-10 inlier, SVHN outlier | AUROC: 0.16 -> 0.83 | Helped |
| Use the white-noise test instead of a single-sided likelihood test | Linear Gaussian outlier detection · CIFAR-10 inlier, SVHN outlier | AUROC: 0.02 -> 0.95 | Helped |
| Use the white-noise test instead of the likelihood-ratio test | AR-DGM outlier detection · TinyImageNet inlier, CIFAR-10 outlier | AUROC: 0.39 -> 0.85 | Helped |
| Use the likelihood-ratio test instead of the white-noise test | AR-DGM outlier detection · CelebA inlier, CIFAR-10 outlier | AUROC: 0.85 -> 0.99 | Hurt |
| Use the white-noise test with VAE residuals rather than the likelihood-ratio test | VAE+Linear outlier detection, nz=64 · TinyImageNet inlier, CIFAR-10 outlier | AUROC: 0.64 -> 0.92 | Helped |
| Increase the VAE latent dimension from nz=64 to nz=512 (ablation) | VAE+Linear white-noise detection · CIFAR-10 inlier, SVHN outlier | AUROC: 0.67 -> 0.88 | Helped |
| Increase the VAE latent dimension from nz=64 to nz=512 (ablation) | VAE+Linear white-noise detection · CIFAR-10 inlier, CelebA outlier | AUROC: 0.64 -> 0.61 | Hurt |
| Use a VAE-based white-noise test rather than AR-DGM or Linear tests | Semantics-oriented CIFAR-10 versus CIFAR-100 benchmark · CIFAR | AUROC: 0.58 -> 0.84 | Helped |
| Apply white-noise testing to VAE residuals rather than a likelihood test | VAE-based outlier detection · CIFAR-10 inlier, SVHN outlier | AUROC: 0.04 -> 0.88 | Helped |
| Use the white-noise test in place of the likelihood-ratio test | AR-DGM outlier detection · Synthetic 2 | AUROC: 0.48 -> 0.76 | Helped |
| Use the likelihood test rather than the white-noise test | AR-DGM outlier detection · Synthetic 3 | AUROC: 0.64 vs 0.57 | Mixed |
Closest papers by meaning
- ODIM: Outlier Detection via Likelihood of Under-Fitted Generative Models (ICML 2024 · similarity 0.91)
- ALTBI: Constructing Improved Outlier Detection Models via Optimization of Inlier-Memorization Effect (AAAI 2025 · similarity 0.90)
- Forte : Finding Outliers with Representation Typicality Estimation (ICLR 2025 · similarity 0.89)
- RODEO: Robust Outlier Detection via Exposing Adaptive Out-of-Distribution Samples (ICML 2024 · similarity 0.89)
- Diffusion-based Semantic Outlier Generation via Nuisance Awareness for Out-of-Distribution Detection (AAAI 2025 · similarity 0.88)
- Systematic Outliers in Large Language Models (ICLR 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.