Research map / Continual learning / Class-incremental
Spectral-Aware Analytic Class-Incremental Learning for Long-Tailed Distributions
Quyen Tran, Ngoc-Hai Nguyen, Minh Quan Dao, Zhuowei Li, Nam Hai, Trung Le, Dimitris N. Metaxas
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 GSR to GACL | GACL (DINO-v2) · Split CIFAR-100 (long-tailed) | Last accuracy: 48.78 -> 65.81 | Improved |
| add GSR to AnaCP | AnaCP (DINO-v2) · Split CIFAR-100 (long-tailed) | Last accuracy: 72.90 -> 82.51 | Improved |
| add GSR to AnaCP | AnaCP (MoCo-v3) · Split CIFAR-100 (long-tailed) | Last accuracy: 57.38 -> 66.01 | Improved |
| add GSR to AnaCP | AnaCP (MoCo-v3) · Split ImageNet-R (long-tailed) | Last accuracy: 36.84 -> 37.21 | No clear effect |
| use linear mixup instead of Standard Ridge (ablation) | GACL · Split CUB-200 (DINO-v2) | Last accuracy: 80.73 -> 77.85 | Hurt |
| replace linear mixup with spherical mixup (ablation) | GACL + mixup · Split CUB-200 (DINO-v2) | Last accuracy: 77.85 -> 82.01 | Improved |
| add adaptive mixing alpha_c (ablation) | GACL + spherical mixup · Split CUB-200 (DINO-v2) | Last accuracy: 82.01 -> 82.46 | Improved |
| add spherical mixup (GSR) on MoCo-v3 (ablation) | GACL · Split CUB-200 (MoCo-v3) | Last accuracy: 34.48 -> 48.33 | Improved |
| add GSR at imbalance ratio 10 | GACL · Split CIFAR-100 rho=10 | Last accuracy: 81.23 -> 85.22 (+3.99) | Improved |
| add GSR at imbalance ratio 150 | GACL · Split CIFAR-100 rho=150 | Last accuracy: 38.60 -> 55.50 (+16.9) | Improved |
| add GSR to GACL, tail classes | GACL · Split CIFAR-100 tail classes | Tail accuracy: 12.50 -> 38.00 | Improved |
| add GSR to GACL, head classes | GACL · Split CIFAR-100 head classes | Head accuracy: 95.38 -> 95.25 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- GACL — GACL: Exemplar-Free Generalized Analytic Continual Learning (NeurIPS 2024) · better on 8 of 8 shared results
- AnaCP — AnaCP: Toward Upper-Bound Continual Learning via Analytic Contrastive Projection (NeurIPS 2025) · better on 5 of 5 shared results
- FeCAM — FeCAM: Exploiting the Heterogeneity of Class Distributions in Exemplar-Free Continual Learning (NeurIPS 2023) · better on 1 of 1 shared result
- RanPAC — RanPAC: Random Projections and Pre-trained Models for 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
- TailAdapt: Heavy-Tailed Sparse Variational Adaptation for Long-Tailed Class Incremental Learning (NeurIPS 2026 · similarity 0.89)
- Confusion-Aware Spectral Regularizer for Long-Tailed Recognition (CVPR 2026 · similarity 0.87)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.86)
- Learning Continually by Spectral Regularization (ICLR 2025 · similarity 0.86)
- Optimal Rates for Vector-Valued Spectral Regularization Learning Algorithms (NeurIPS 2024 · similarity 0.85)
- Continuous Contrastive Learning for Long-Tailed Semi-Supervised Recognition (NeurIPS 2024 · similarity 0.85)
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.
