Research map / Continual learning / Class-incremental
Continual Out-of-Distribution Detection with Analytic Neural Collapse
Saleh Momeni, Changnan Xiao, Bing Liu
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 class-variation collapse (NC1) on top of random projection (ablation) | AnaNC · CUB in-dataset OOD | AUC: 70.61 -> 78.56 | Improved |
| add simplex ETF arrangement (NC2) on top of NC1 (ablation) | AnaNC · CUB in-dataset OOD | AUC: 78.56 -> 83.07 | Improved |
| add simplex ETF arrangement (NC2) on top of NC1 (ablation) | AnaNC · Cars in-dataset OOD | AUC: 69.52 -> 81.76 | Improved |
| add simplex ETF arrangement (NC2) on top of NC1 (ablation) | AnaNC · Cars in-dataset OOD | FPR95: 87.61 -> 76.13 | Improved |
| add NC1 on top of random projection (ablation) | AnaNC · CIFAR100 in-dataset OOD | AUC: 85.09 -> 86.30 | Improved |
| add random projection alone to NCM on PTM features (ablation) | NCM · CUB in-dataset OOD | AUC: 71.18 -> 70.61 | No clear effect |
| add random projection alone to NCM on PTM features (ablation) | NCM · Cars in-dataset OOD | AUC: 58.68 -> 59.14 | No clear effect |
| increase random projection dimension D (ablation) | AnaNC · In-dataset OOD, DINO | AUC | Improved |
| use AnaNC instead of best baseline (in-dataset, DINO) | AnaNC · In-dataset OOD average | AUC: 2.22% in AUC and 3.24% in FPR95 | Improved |
| use AnaNC instead of best baseline (in-dataset, MOCO) | AnaNC · In-dataset OOD average | AUC: 1.51% in AUC and 4.38% in FPR95 | Improved |
| use AnaNC instead of strongest baseline (cross-dataset, DINO) | AnaNC · Cross-dataset OOD average | AUC: 0.49% in AUC and 0.88% in FPR95 | Improved |
| use AnaNC instead of strongest baseline (cross-dataset, MOCO) | AnaNC · Cross-dataset OOD average | AUC: 0.86% in AUC and 1.94% in FPR95 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- RanPac — RanPAC: Random Projections and Pre-trained Models for Continual Learning (NeurIPS 2023) · better on 8 of 8 shared results
- SLCA — SLCA: Slow Learner with Classifier Alignment for Continual Learning on a Pre-trained Model (ICCV 2023) · better on 2 of 2 shared results
- FECAM — FeCAM: Exploiting the Heterogeneity of Class Distributions in Exemplar-Free Continual Learning (NeurIPS 2023) · better on 1 of 1 shared result
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
- Towards Understanding Catastrophic Forgetting in Two-layer Convolutional Neural Networks (ICML 2025 · similarity 0.88)
- Controlling Neural Collapse Enhances Out-of-Distribution Detection and Transfer Learning (ICML 2025 · similarity 0.88)
- NECO: NEural Collapse Based Out-of-distribution detection (ICLR 2024 · similarity 0.88)
- Detecting Out-of-Distribution Through the Lens of Neural Collapse (CVPR 2025 · similarity 0.87)
- Pursuing Feature Separation based on Neural Collapse for Out-of-Distribution Detection (ICLR 2025 · similarity 0.87)
- Forgetting, Ignorance or Myopia: Revisiting Key Challenges in Online Continual Learning (NeurIPS 2024 · 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.
