Research map / Video / Video generation
Scalable RF Simulation in Generative 4D Worlds
Zhiwei Zheng, Dongyin Hu, Mingmin Zhao
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 path masking (ablation) | state-aware causal transformer · motion generation | FID: 0.747 -> 0.238 | Improved |
| add path masking (ablation) | state-aware causal transformer · motion generation | R-Precision: 0.643 -> 0.755 | Improved |
| add spatial state information (ablation) | state-aware causal transformer · motion generation | mean path error: 0.250 -> 0.151 | Improved |
| add spatial state information (ablation) | state-aware causal transformer · motion generation | mean ending error: 0.460 -> 0.287 | Improved |
| use masking rate [0.5, 0.9] instead of [0.1, 0.5] (ablation) | path masking · motion generation | FID: 0.396 -> 0.238 | Improved |
| use masking rate [0.5, 0.9] instead of [0.1, 0.9] (ablation) | path masking · motion generation | FID: 0.298 -> 0.238 | Improved |
| lengthen contiguous masking segment from 5 to 15 points (ablation) | path masking · motion generation | R-Precision: 0.755 -> 0.776 | Improved |
| lengthen contiguous masking segment from 5 to 15 points (ablation) | path masking · motion generation | mean path error: 0.151 -> 0.228 | Hurt |
| pool path tokens into one token (mean) instead of keeping all tokens (ablation) | state-aware causal transformer · motion generation | mean path error: 0.151 -> 0.201 | Hurt |
| add WaveVerse simulated data (4x) to real training data | RF imaging model · RF imaging | MAE: 20.10 -> 18.08 | Improved |
| add Standard RT simulated data (4x) to real training data | RF imaging model · RF imaging | MAE: 20.10 -> 22.28 | Hurt |
| add WaveVerse simulated data (19x) to real training data | activity classifier · human activity recognition | accuracy: 31.6% -> 71.6% | Improved |
Closest papers by meaning
- Radio Frequency Ray Tracing with Neural Object Representation for Enhanced RF Modeling (CVPR 2025 · similarity 0.85)
- GSRF: Complex-Valued 3D Gaussian Splatting for Efficient Radio-Frequency Data Synthesis (NeurIPS 2025 · similarity 0.83)
- RxGS: Receiver-Generalizable 3D Gaussian Splatting for Radio-Frequency Data Synthesis (NeurIPS 2026 · similarity 0.83)
- Differentiable and Learnable Wireless Simulation with Geometric Transformers (ICLR 2025 · similarity 0.82)
- MITO: A Millimeter-Wave Dataset and Simulator for Non-Line-of-Sight Perception (NeurIPS 2026 · similarity 0.82)
- Can NeRFs "See" without Cameras? (NeurIPS 2025 · similarity 0.81)
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.
