Research map / Federated learning and privacy / Personalized and heterogeneous FL
A Unified Federated Framework for Trajectory Data Preparation via LLMs
Zhihao Zeng, Ziquan Fang, Wei Shao, Lu Chen, Yunjun Gao
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 TKE (trajectory knowledge enhancer) (ablation) | FedTDP · 10 TDP tasks | overall performance: at least 27.52% degradation without TKE | Improved |
| add TPA (privacy autoencoder) (ablation) | FedTDP · 10 TDP tasks | overall performance: approximately 5GB additional communication over 10 rounds | Hurt |
| add FPO (federated parallel optimization) (ablation) | FedTDP · 10 TDP tasks | runtime and communication: almost 4 times less | Improved |
| use FedTDP instead of single-task methods | S-TDP · Table 2 tasks | F1/accuracy: at least 18.38% improvement | Improved |
| use FedTDP instead of tabular LLM methods | FM4DP, MELD, TableGPT · Table 2 tasks | F1/accuracy: at least 32.26% | Improved |
| use FedTDP instead of spatio-temporal LLM methods | PromptGAT, UniST, UrbanGPT · Table 2 tasks | F1/accuracy: 4.84% to 45.22% | Improved |
| train on more tasks (multi-task) (ablation) | FedTDP · 10 TDP tasks | performance: falls below S-TDP in Table 2 when trained on one task | Improved |
| use FedTDP framework for training | LLM-based baselines · training efficiency | runtime: reduced by a factor of 11.3 to 14.2 | Improved |
| use FedTDP at test time | LLM-based baselines · testing efficiency | runtime: 2.6 to 4.8 times lower | Improved |
| transmit only data needed for cross-client TDP | FedTDP · testing efficiency | communication size: 1.4 to 1.8 times less | Improved |
Closest papers by meaning
- PLMTrajRec: A Scalable and Generalizable Trajectory Recovery Method with Pre-trained Language Models (NeurIPS 2025 · similarity 0.85)
- TrajAgent: An LLM-Agent Framework for Trajectory Modeling via Large-and-Small Model Collaboration (NeurIPS 2025 · similarity 0.84)
- Pedestrian Trajectory Prediction with Missing Data: Datasets, Imputation, and Benchmarking (NeurIPS 2024 · similarity 0.84)
- C2F-TP: A Coarse-to-Fine Denoising Framework for Uncertainty-Aware Trajectory Prediction (AAAI 2025 · similarity 0.84)
- TrajMamba: An Efficient and Semantic-rich Vehicle Trajectory Pre-training Model (NeurIPS 2025 · similarity 0.84)
- ProDiff: Prototype-Guided Diffusion for Minimal Information Trajectory Imputation (ICML 2025 · similarity 0.83)
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.
