Research map / Probabilistic and scientific modeling / Variational and generative sampling
Likelihood Regret: An Out-of-Distribution Detection Score For Variational Auto-encoder
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 Likelihood Regret score instead of raw likelihood | VAE trained on CIFAR-10 · OOD SVHN | AUCROC: 0.193 -> 0.875 | Helped |
| use Likelihood Regret instead of raw likelihood | VAE trained on Fashion-MNIST · OOD MNIST | AUCROC: 0.201 -> 0.988 | Helped |
| Likelihood Regret vs input-complexity adjusted score (png) | VAE trained on CIFAR-10 · OOD Noise | AUCROC: 0.032 -> 0.994 | Helped |
| Likelihood Regret vs likelihood ratio with background model | VAE trained on CIFAR-10 · OOD SVHN | AUCROC: 0.265 -> 0.875 | Helped |
| Likelihood Regret vs input complexity (png) on hard-ish OOD set | VAE trained on CIFAR-10 · OOD SVHN | AUCROC: 0.912 -> 0.875 | Hurt |
| optimize whole VAE (encoder and decoder) instead of only encoder when computing regret (ablation) | Likelihood Regret · Fashion-MNIST VAE, OOD MNIST | AUCROC: 0.988 -> 0.124 | Hurt |
| optimize decoder only instead of encoder when computing regret (ablation) | Likelihood Regret · CIFAR-10 VAE, OOD CelebA | AUCROC: 0.655 -> 0.714 | Helped |
| optimize whole encoder (LR_E) instead of only posterior mean and variance (LR_Z) (ablation) | Likelihood Regret · Fashion-MNIST VAE, OOD MNIST | AUCROC: 0.967 -> 0.988 | Helped |
| vary beta in beta-VAE (0.1 to 10) (ablation) | Likelihood Regret · Fashion-MNIST and CIFAR-10 VAEs | AUCROC | No clear effect |
| increase VAE capacity 4x (ablation) | Likelihood Regret · CIFAR-10 VAE, OOD SVHN | AUCROC: 0.875 -> 0.79 | Mixed |
Closest papers by meaning
- A Geometric Explanation of the Likelihood OOD Detection Paradox (ICML 2024 · similarity 0.90)
- Forte : Finding Outliers with Representation Typicality Estimation (ICLR 2025 · similarity 0.88)
- The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection (NeurIPS 2024 · similarity 0.88)
- Understanding Likelihood of Normalizing Flow and Image Complexity through the Lens of Out-of-Distribution Detection (AAAI 2024 · similarity 0.87)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 2025 · similarity 0.87)
- Out-of-Distribution Detection with Prototypical Outlier Proxy (AAAI 2025 · 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.