Research map / Time series / Foundation models for time series
Mobility Data Representations for Spatiotemporal Tasks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
A PhD research summary on developing interpretable, reusable, and geographically transferable trajectory (mobility data) representations, covering TIF, Geolet, and ongoing shapelet-based and map-matching work.
Problem
Mobility data analysis lacks reusable, transferable, and interpretable trajectory representations; models often fail to transfer across geographic regions and public data is scarce due to privacy constraints.
Approach
Transform trajectories into tabular or shapelet-based representations (TIF with human-defined mathematical features; Geolet with spatiotemporal shapelets and normalization of subsequences) to enable interpretable use with standard ML models and improve geographic transferability.
Key results
- Geolet reportedly shows performance comparable to SotA methods on classification tasks while offering improved interpretability, though it lags in runtime.
- In map-matching study, given a shapelet, the candidate road space can be reduced by up to 90% with over 85% probability that the correct road remains among the remaining candidates.
- TIF showed extremely good performance across multiple public benchmark datasets (no specific numbers given).
Limitations
- TIF relies on a finite set of human-defined features that may not suit specific tasks.
- Geolet lags in runtime due to non-linear similarity computation.
- Geographic transferability of learned models remains a challenge, with preliminary results only.
Experimental setup
Datasets: GPS traces from OpenStreetMap Eclipse SUMO simulated trips for Rome, New York, Athens
Baselines: Trajectory Interval Forest
Builds on: Canonical Interval Forest shapelets
Closest papers by meaning
- TraXion: Rethinking Pre-training Frameworks for Mobility and Beyond (NeurIPS 2026 · similarity 0.80)
- TransferTraj: A Vehicle Trajectory Learning Model for Region and Task Transferability (NeurIPS 2025 · similarity 0.79)
- Satellites Reveal Mobility: A Commuting Origin-destination Flow Generator for Global Cities (NeurIPS 2025 · similarity 0.79)
- Generative Human Trajectory Recovery via Embedding-Space Conditional Diffusion (ICML 2025 · similarity 0.79)
- Holistic Semantic Representation for Navigational Trajectory Generation (AAAI 2025 · similarity 0.79)
- GTG: Generalizable Trajectory Generation Model for Urban Mobility (AAAI 2025 · similarity 0.79)
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.