Research map / Label-efficient and robust learning / Recognition and long-tail
Bridging Class Imbalance and Partial Labeling via Spectral-Balanced Energy Propagation for Skeleton-based Action Recognition
Yandan Wang, Chenqi Guo, Yinglong Ma, Jiangyan Chen, Yuan Gao, Weiming Dong
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 spectral balancing and energy-based tightened reliability to label propagation (ablation) | SpeLER · NTU-60 X-sub (pi=30, 25% labels) | top-1 accuracy: 63.30 -> 65.02 | Improved |
| add spectral balancing (SpeL) (ablation) | SpeLER · NTU-60 X-sub (pi=30, 25% labels) | top-1 accuracy: 64.24 -> 65.02 | Improved |
| add energy-based tightened reliability (ER) (ablation) | SpeLER · NTU-60 X-sub (pi=30, 25% labels) | top-1 accuracy: 64.36 -> 65.02 | Improved |
| use label propagation instead of FixMatch consistency (ablation) | SpeLER · NTU-60 X-sub (pi=30, 25% labels) | top-1 accuracy: 64.39 -> 65.02 | Improved |
| add time-frequency cross-domain fusion (ablation) | SpeLER · NTU-60 X-sub (pi=30, 25% labels) | top-1 accuracy: 64.19 -> 65.02 | Improved |
| add time-frequency cross-domain fusion (ablation) | SpeLER · SleepEDF (10% labels) | accuracy: 82.26 -> 84.31 | Improved |
| adapt FreqMix (time-frequency modelling) to the semi-supervised imbalanced setting | FreqMix. · NTU-60 X-sub (25% labels, pi=60) | top-1 accuracy: 59.19 vs 65.02 (SpeLER) | Hurt |
| SpeLER at highly imbalanced 50% labels (pi=100) versus protoLP | SpeLER · NTU-60 X-sub (50% labels, pi=100) | top-1 accuracy: 66.49 (protoLP) vs 66.45 | No clear effect |
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.
- SCD-Net — SCD-Net: Spatiotemporal Clues Disentanglement Network for Self-Supervised Skeleton-Based Action Recognition (AAAI 2024) · better on 5 of 5 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
- SA-DVAE: Improving Zero-Shot Skeleton-Based Action Recognition by Disentangled Variational Autoencoders (ECCV 2024 · similarity 0.85)
- Skeleton-based Action Recognition with Non-linear Dependency Modeling and Hilbert-Schmidt Independence Criterion (AAAI 2025 · similarity 0.84)
- Ske2Grid: Skeleton-to-Grid Representation Learning for Action Recognition (ICML 2023 · similarity 0.84)
- Balanced Data, Imbalanced Spectra: Unveiling Class Disparities with Spectral Imbalance (ICML 2024 · similarity 0.84)
- Robust Visual Recognition with Class-Imbalanced Open-World Noisy Data (AAAI 2024 · similarity 0.83)
- MaskCLR: Attention-Guided Contrastive Learning for Robust Action Representation Learning (CVPR 2024 · similarity 0.83)
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.
