Research map / Probabilistic and scientific modeling / Dynamical systems
VIBE: Disentangling Social Dynamics via Kinematics-Informed Variational Inference for Behavioral Emotion
Abhishek Pratap Singh, Vaibhav Pratap Singh, Deepak Kumar, Balasubramanian Raman
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 |
|---|---|---|---|
| use synchrony-gated decoupled AdaLN instead of standard AdaLN (ablation) | VIBE · VGAF | accuracy: 68.20% -> 70.17% | Improved |
| use synchrony-gated decoupled AdaLN instead of standard AdaLN (ablation) | VIBE · GECV | accuracy: 91.82% -> 91.85% | No clear effect |
| use decoupled AdaLN instead of naive concatenation (ablation) | VIBE · VGAF | accuracy: 66.52% -> 70.17% | Improved |
| add orthogonality and KL losses to semantic alignment loss (ablation) | VIBE · VGAF | accuracy: 68.41% -> 70.17% | Improved |
| add orthogonality and KL losses to semantic alignment loss (ablation) | VIBE · GECV | accuracy: 90.18% -> 91.85% | Improved |
| add L_ortho and L_KL without semantic alignment (ablation) | VIBE · VGAF | accuracy: 67.33% -> 66.70% | Hurt |
| add L_ortho and L_KL without semantic alignment (ablation) | VIBE · GECV | accuracy: 90.97% -> 89.39% | Hurt |
| remove orthogonality loss (keep KL and alignment) (ablation) | VIBE · GECV | accuracy: 91.85% -> 87.74% | Hurt |
| increase tracked agents K from 8 to 10 (ablation) | VIBE · VGAF | accuracy: 70.17% -> 68.88% | Hurt |
| increase tracked agents K from 8 to 10 (ablation) | VIBE · GECV | accuracy: 91.85% -> 91.82% | No clear effect |
| reduce tracked agents K from 8 to 3 (ablation) | VIBE · VGAF | accuracy: 70.17% -> 67.80% | Hurt |
Closest papers by meaning
- EmotiCon: Context-Aware Multimodal Emotion Recognition Using Frege's Principle (CVPR 2020 · similarity 0.86)
- Resonance: Learning to Predict Social-Aware Pedestrian Trajectories as Co-Vibrations (ICCV 2025 · similarity 0.86)
- AffectGPT: A New Dataset, Model, and Benchmark for Emotion Understanding with Multimodal Large Language Models (ICML 2025 · similarity 0.86)
- $E^3$: Exploring Embodied Emotion Through A Large-Scale Egocentric Video Dataset (NeurIPS 2024 · similarity 0.86)
- Behavioral Geometric Supervision Aligns Video Foundation Models with Human Social Perception (NeurIPS 2026 · similarity 0.85)
- VibeCheck: Discover and Quantify Qualitative Differences in Large Language Models (ICLR 2025 · 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.
