Research map / Label-efficient and robust learning / OOD and anomaly detection
Anomaly Detection With Multiple-Hypotheses Predictions
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 ConAD with 8 hypotheses instead of the strongest reported CIFAR-10 baseline | CIFAR-10 anomaly detection · CIFAR-10 one-vs-nine tasks | AUROC: 63.2 -> 67.1 | Helped |
| Use ConAD with 2 hypotheses instead of MHP with 2 hypotheses | Metal Anomaly anomaly detection · Metal Anomaly | AUROC: 98.0 (0.5) -> 98.5 (0.1) | Helped |
| Add a discriminator (ablation) | ConAD (8 hypotheses) · CIFAR-10 one-vs-nine tasks | AUROC: 61.9 -> 67.1 | Helped |
| Add winner-takes-all loss (ablation) | ConAD (8 hypotheses) · CIFAR-10 one-vs-nine tasks | AUROC: 61.8 -> 67.1 | Helped |
| Add WTA and avoid loose hypothesis coupling (ablation) | ConAD (8 hypotheses) · CIFAR-10 one-vs-nine tasks | AUROC: 61.0 -> 67.1 | Helped |
| Add multiple hypotheses (ablation) | ConAD · CIFAR-10 one-vs-nine tasks | AUROC: 61.7 -> 67.1 | Helped |
| Add multiple hypotheses and a discriminator (ablation) | ConAD · CIFAR-10 one-vs-nine tasks | AUROC: 61.0 -> 67.1 | Helped |
| Use 8 rather than 2 hypotheses (ablation) | ConAD · CIFAR-10 one-vs-nine tasks | AUROC: 64.3 -> 67.1 | Helped |
| Increase the hypothesis count from 8 to 16 (ablation) | ConAD · CIFAR-10 one-vs-nine tasks | AUROC: 67.1 -> 65.9 | Hurt |
| Increase the hypothesis count from 2 to 8 (ablation) | ConAD · Metal Anomaly | AUROC: 98.5 (0.1) -> 96.5 (0.2) | Mixed |
Closest papers by meaning
- Understanding Anomaly Detection with Deep Invertible Networks through Hierarchies of Distributions and Features (NeurIPS 2020 · similarity 0.88)
- Two-Layer Convolutional Autoencoders Trained on Normal Data Provably Detect Unseen Anomalies (ICLR 2026 · similarity 0.87)
- ResAD: A Simple Framework for Class Generalizable Anomaly Detection (NeurIPS 2024 · similarity 0.86)
- CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection (AAAI 2025 · similarity 0.85)
- AnomalyNCD: Towards Novel Anomaly Class Discovery in Industrial Scenarios (CVPR 2025 · similarity 0.85)
- Beyond Normal References: Discriminative Few-Shot Anomaly Detection (NeurIPS 2026 · similarity 0.84)
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.