Research map / Label-efficient and robust learning / OOD and anomaly detection
Unleashing Mask: Explore the Intrinsic Out-of-Distribution Detection Capability
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 |
|---|---|---|---|
| Apply UM to Energy | Energy · CIFAR-10 | FPR95: 42.69\pm 1.31 -> 33.29\pm 1.70 | Helped |
| Apply UMAP to Energy | Energy · CIFAR-10 | FPR95: 42.69\pm 1.31 -> 30.71\pm\textbf{1.94} | Helped |
| Apply UM to OE | OE · CIFAR-10 | FPR95: 13.80\pm 0.28 -> 11.22\pm \textbf{0.16} | Helped |
| Apply UMAP to OE | OE · CIFAR-10 | FPR95: 13.80\pm 0.28 -> 12.21\pm0.09 | Helped |
| Apply UM to Energy | Energy · CIFAR-100 | FPR95: 78.46\pm 0.06 -> 74.05\pm 0.55 | Helped |
| Apply UMAP to Energy | Energy · CIFAR-100 | FPR95: 78.46\pm 0.06 -> \textbf{72.21}\pm\textbf{1.46} | Helped |
| Apply UM to OE | OE · CIFAR-100 | FPR95: 34.73\pm 3.85 -> 34.71\pm0.81 | No clear effect |
| Apply UMAP to OE | OE · CIFAR-100 | FPR95: 34.73\pm 3.85 -> \textbf{33.62}\pm\textbf{0.26} | Helped |
| Use mask ratios from 0.995 to 0.97 to mine atypical samples (ablation) | UM · OOD detection benchmarks | FPR95 | Helped |
| Learn a pruning mask with UMAP rather than use vanilla pruning with its original objective (ablation) | UMAP · OOD detection benchmarks | FPR95 | Helped |
| Use pruning rates across the reported range (ablation) | UMAP · OOD detection benchmarks | FPR95 | No clear effect |
| Use UM or UMAP instead of conventional dropout, weight decay, or early stopping to target OOD forgetting (ablation) | OOD detection model · CIFAR-10 | FPR95 | Helped |
Closest papers by meaning
- Splitting & Integrating: Out-of-Distribution Detection via Adversarial Gradient Attribution (ICML 2025 · similarity 0.90)
- What If the Input is Expanded in OOD Detection? (NeurIPS 2024 · similarity 0.89)
- Can We Ignore Labels in Out of Distribution Detection? (ICLR 2025 · similarity 0.89)
- Beyond Clean Training Data: A Versatile and Model-Agnostic Framework for Out-of-Distribution Detection with Contaminated Training Data (CVPR 2025 · similarity 0.89)
- The Best of Both Worlds: On the Dilemma of Out-of-distribution Detection (NeurIPS 2024 · similarity 0.88)
- NECO: NEural Collapse Based Out-of-distribution detection (ICLR 2024 · 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.