Research map / Label-efficient and robust learning / Recognition and long-tail
PREDICT & CLUSTER: Unsupervised Skeleton Based Action Recognition
Check 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 |
|---|---|---|---|
| Train the encoder using a weakened Fixed Weights decoder rather than leave it random and untrained (ablation) | P&C · NW-UCLA | accuracy: 72.0 -> 84.9 | Helped |
| Train the encoder using a weakened Fixed States decoder rather than leave it random and untrained (ablation) | P&C · NW-UCLA | accuracy: 72.0 -> 83.8 | Helped |
| Use Fixed Weights decoder rather than Fixed States decoder (ablation) | P&C · NW-UCLA | accuracy: 83.8 -> 84.9 | Mixed |
| Train with Fixed Weights decoder rather than leave the encoder random and untrained (ablation) | P&C · UWA3D V3 | accuracy: 48.5 -> 59.9 | Helped |
| Train with Fixed States decoder rather than leave the encoder random and untrained (ablation) | P&C · UWA3D V3 | accuracy: 48.5 -> 59.5 | Helped |
| Train with Fixed Weights decoder rather than leave the encoder random and untrained (ablation) | P&C · UWA3D V4 | accuracy: 51.5 -> 63.1 | Helped |
| Train with Fixed States decoder rather than leave the encoder random and untrained (ablation) | P&C · UWA3D V4 | accuracy: 51.5 -> 63.1 | Helped |
| Train with Fixed States decoder rather than leave the encoder random and untrained (ablation) | P&C · NTU RGB+D 60 Cross-View | accuracy: 56.4 -> 76.3 | Helped |
| Train with Fixed Weights decoder rather than leave the encoder random and untrained (ablation) | P&C · NTU RGB+D 60 Cross-View | accuracy: 56.4 -> 76.1 | Helped |
| Train with Fixed States decoder rather than leave the encoder random and untrained (ablation) | P&C · NTU RGB+D 60 Cross-Subject | accuracy: 39.6 -> 50.6 | Helped |
| Train with Fixed Weights decoder rather than leave the encoder random and untrained (ablation) | P&C · NTU RGB+D 60 Cross-Subject | accuracy: 39.6 -> 50.7 | Helped |
| Use a weakened decoder (Fixed Weights or Fixed States) during training (ablation) | P&C · all evaluated datasets | accuracy | Helped |
Closest papers by meaning
- Unsupervised Representation Learning With Long-Term Dynamics for Skeleton Based Action Recognition (AAAI 2018 · similarity 0.91)
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (CVPR 2025 · similarity 0.90)
- Heterogeneous Skeleton-Based Action Representation Learning (CVPR 2025 · similarity 0.89)
- Neuron: Learning Context-Aware Evolving Representations for Zero-Shot Skeleton Action Recognition (CVPR 2025 · similarity 0.89)
- Stitch, Contrast, and Segment: Learning a Human Action Segmentation Model Using Trimmed Skeleton Videos (AAAI 2025 · similarity 0.88)
- Recovering Complete Actions for Cross-dataset Skeleton Action Recognition (NeurIPS 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.