Research map / Multi-agent systems and games / Human-AI and goal-conditioned
HyperDynamics: Meta-Learning Object and Agent Dynamics with Hypernetworks
Zhou Xian, Shamit Lal, Hsiao-Yu Tung, Emmanouil Antonios Platanios, Katerina Fragkiadaki
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 |
|---|---|---|---|
| Generate system-conditioned dynamics experts instead of using a fixed global model | HyperDynamics vs. Direct · Object pushing | prediction error | Improved |
| Generate system-conditioned dynamics experts instead of using a fixed global model | HyperDynamics vs. Direct · Object pushing | success rate | Improved |
| Infer system properties and generate an expert for unseen objects | HyperDynamics vs. XYZ · Unseen object pushing | success rate | Improved |
| Generate dynamics directly from inferred properties rather than adapt parameters with gradient steps | HyperDynamics vs. MB-MAML · Object pushing with 5 interaction samples | prediction error | Improved |
| Generate system-conditioned dynamics experts | HyperDynamics vs. Recurrent and MB-MAML · Locomotion | average return | Improved |
| Generate system-conditioned dynamics experts | HyperDynamics vs. Recurrent and MB-MAML · Locomotion prediction error | mean squared error | Improved |
| Add top-down shape reconstruction to regularize training (ablation) | HyperDynamics for object pushing · Object pushing | prediction error | Improved |
| Crop the scene feature map to produce object-centric features (ablation) | HyperDynamics for object pushing · Object pushing | prediction error | Improved |
| Use detected rather than ground-truth object positions (ablation) | HyperDynamics for object pushing · Object pushing | prediction error | Hurt |
| Increase the visual latent code dimension beyond the chosen size (ablation) | HyperDynamics for object pushing · Object pushing | prediction error | No clear effect |
| Increase the interaction latent code dimension beyond the chosen size (ablation) | HyperDynamics · Object pushing and locomotion | prediction error | No clear effect |
Closest papers by meaning
- Meta-Dynamical State Space Models for Integrative Neural Data Analysis (ICLR 2025 · similarity 0.87)
- Learning on the Go: A Meta-learning Object Navigation Model (ICCV 2025 · similarity 0.87)
- InterDyn: Controllable Interactive Dynamics with Video Diffusion Models (CVPR 2025 · similarity 0.87)
- Efficient Dynamics Modeling in Interactive Environments with Koopman Theory (ICLR 2024 · similarity 0.86)
- Neural Persistence Dynamics (NeurIPS 2024 · similarity 0.85)
- Dyn-O: Building Structured World Models with Object-Centric Representations (NeurIPS 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.
