Research map / Learning theory and algorithms / Matrix and tensor methods
Efficient Temporal Sequence Comparison and Classification Using Gram Matrix Embeddings on a Riemannian Manifold
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 |
|---|---|---|---|
| Embed temporal sequences as regularized Hankel Gram matrices and compare them with a PD-manifold metric | 3D action recognition · MSR-Action3D, protocol [36] | accuracy (%): 91.50 -> 94.74 | Helped |
| Embed temporal sequences as regularized Hankel Gram matrices and compare them with a PD-manifold metric | 3D action recognition · MSR-Action3D, protocol [29] | accuracy (%): 85.16±3.13 -> 90.35±2.66 | Helped |
| Use Gram-matrix embedding with KLDM | 3D action recognition · MHAD | accuracy (%): 100 -> 100 | No clear effect |
| Embed sequences as Gram matrices and classify using a manifold metric | 3D action recognition · UTKinect-Action | average accuracy: 91.5 -> 100 | Helped |
| Embed sequences as Gram matrices and classify using a manifold metric | 3D action recognition · HDM05 | accuracy (%): 79.1 ± 7.5 -> 88.0 ± 6.3 | Helped |
| Increase Hankel block row size from r = 1 to r = 9 (ablation) | G-J · MSR-Action3D, AS1/AS2/AS3 protocol | average accuracy (%): 89.73 -> 96.97 | Helped |
| Increase Hankel block row size from r = 1 to r = 12 (ablation) | G-A · MSR-Action3D, AS1/AS2/AS3 protocol | average accuracy (%): 89.47 -> 96.96 | Helped |
| Use G-K rather than G-A (ablation) | Gram-matrix embedding · MSR-Action3D, AS1/AS2/AS3 protocol | average accuracy (%): 89.47 vs 87.53; 96.96 vs 95.80 | Mixed |
| Use G-J rather than G-K (ablation) | Gram-matrix embedding · MSR-Action3D, AS1/AS2/AS3 protocol | average accuracy (%): 89.73 vs 87.53; 96.97 vs 95.98 | Mixed |
Closest papers by meaning
- Matrix Manifold Neural Networks++ (ICLR 2024 · similarity 0.85)
- Learned Trajectory Embedding for Subspace Clustering (CVPR 2024 · similarity 0.84)
- Task-Induced Riemannian Metrics for Vision Transformer Feature Spaces (NeurIPS 2026 · similarity 0.84)
- Geometry-Aware Similarity Metrics for Neural Representations on Riemannian and Statistical Manifolds (NeurIPS 2026 · similarity 0.84)
- Connecting Neural Models Latent Geometries with Relative Geodesic Representations (NeurIPS 2025 · similarity 0.84)
- RMLR: Extending Multinomial Logistic Regression into General Geometries (NeurIPS 2024 · similarity 0.84)
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.