Research map / Video / Video-language understanding
SLAP: The Semantic Least Action Principle for Variational Video-Language Modeling
Xiang Fang, Wanlong Fang
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 kinetic energy term (mu>0) (ablation) | SLAP · Tunnel Test | accuracy (%): 62.0% -> 83.9% | Improved |
| add kinetic energy term (mu>0) (ablation) | SLAP · MSR-VTT (10% frames) | accuracy (%): 38.5% -> 41.8% | Improved |
| add learned potential energy term (ablation) | SLAP · Tunnel Test | accuracy (%): 41.5% -> 83.9% | Improved |
| add learned potential energy term (ablation) | SLAP · MSR-VTT (10% frames) | accuracy (%): 40.1% -> 41.8% | Improved |
| use learned potential instead of static cosine potential (ablation) | SLAP · Tunnel Test | accuracy (%): 70.5% -> 83.9% | Improved |
| use learned potential instead of static cosine potential (ablation) | SLAP · MSR-VTT (10% frames) | accuracy (%): 42.1% -> 41.8% | Hurt |
| use geodesic kinetic term instead of L2 kinetic (ablation) | SLAP · Tunnel Test | accuracy (%): 78.2% -> 83.9% | Improved |
| use geodesic kinetic term instead of L2 kinetic (ablation) | SLAP · MSR-VTT (10% frames) | accuracy (%): 41.0% -> 41.8% | Improved |
| raise coupling lambda from 0.5 to 5.0 (ablation) | SLAP · Tunnel Test | accuracy (%): 83.9% -> 52.4% | Hurt |
| raise coupling lambda from 0.5 to 1.0 (ablation) | SLAP · Tunnel Test | accuracy (%): 83.9% -> 79.1% | Hurt |
| replace pure-inertia/linear trajectory with SLAP action minimization | SLERP (Linear) · Tunnel Test | accuracy (%): 41.5% -> 83.9% | Improved |
| use SLAP instead of Video-LLaMA 3 when 90% of frames are dropped | Video-LLaMA 3 · MSR-VTT (10% frames) | accuracy (%): 34.7 -> 41.8 | Improved |
Closest papers by meaning
- Disen-Forcing: Disentangling Semantic Anchoring from Motion for Autoregressive Video Diffusion (NeurIPS 2026 · similarity 0.88)
- Dynamics: Language-Based Representation for Inferring Rigid-Body Dynamics From Videos (CVPR 2026 · similarity 0.87)
- Predictive Structure Improves Video Diffusion Dynamics (ECCV 2026 · similarity 0.87)
- MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models (CVPR 2025 · similarity 0.87)
- Bernini: Latent Semantic Planning for Video Diffusion (NeurIPS 2026 · similarity 0.86)
- Auteur: Language-Driven Cinematographic Framing for Human-Centric Video Generation (NeurIPS 2026 · similarity 0.86)
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.
