Research map / Label-efficient and robust learning / OOD and anomaly detection
Gradient Short-Circuit: Efficient Out-of-Distribution Detection via Feature Intervention
Jiawei Gu, Ziyue Qiao, Zechao Li
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 |
|---|---|---|---|
| add gradient short-circuit to ConjNorm-level detection (GSC vs ConjNorm) | DenseNet-101 · CIFAR-10 | FPR95: 13.92 ± 0.27 -> 7.91 ± 0.18 | Improved |
| GSC vs ConjNorm | DenseNet-101 · CIFAR-100 | FPR95: 28.27 ± 0.44 -> 23.15 ± 0.35 | Improved |
| add ASH activation scaling on top of GSC | GSC · CIFAR-10 | FPR95: 7.91 ± 0.18 -> 10.62 ± 0.19 | Hurt |
| add ASH activation scaling on top of GSC | GSC · CIFAR-100 | FPR95: 23.15 ± 0.35 -> 25.75 ± 0.38 | Hurt |
| GSC vs ConjNorm | ResNet-50 · ImageNet-1K Places365 | FPR95: 62.07 ± 0.65 -> 43.74 ± 0.61 | Improved |
| GSC vs ConjNorm | ResNet-50 · ImageNet-1K iNaturalist | FPR95: 9.62 ± 0.19 -> 11.11 ± 0.15 | Hurt |
| GSC vs ConjNorm | ViT-B/16 · ImageNet-1K avg | FPR95: 36.97 -> 33.62 | Improved |
| GSC vs ConjNorm | Swin-B · ImageNet-1K avg | FPR95: 34.28 -> 31.26 | Improved |
| use zero-out instead of small perturbation (ablation) | GSC · CIFAR-100 (5% mask) | FPR95: 28.64 -> 25.75 | Improved |
| use zero-out instead of orthogonal projection (ablation) | GSC · CIFAR-100 (5% mask) | FPR95: 29.32 -> 25.75 | Improved |
| raise mask ratio from 5% to 10% (ablation) | GSC Zero · CIFAR-100 | FPR95: 25.75 -> 24.10 | Improved |
| add first-order approximation (ablation) | GSC · CIFAR-100 inference | relative time: 3.78 ± 0.12 -> 1.98 ± 0.05 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ReAct — ReAct: Out-of-distribution Detection With Rectified Activations (NeurIPS 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Splitting & Integrating: Out-of-Distribution Detection via Adversarial Gradient Attribution (ICML 2025 · similarity 0.92)
- Detecting Out-of-Distribution Through the Lens of Neural Collapse (CVPR 2025 · similarity 0.91)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 2025 · similarity 0.91)
- Fast Decision Boundary based Out-of-Distribution Detector (ICML 2024 · similarity 0.90)
- Mysteries of the Deep: Role of Intermediate Representations in Out of Distribution Detection (NeurIPS 2025 · similarity 0.90)
- Revisiting Logit Distributions for Reliable Out-of-Distribution Detection (NeurIPS 2025 · similarity 0.90)
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.
