GeoSplit reduces map data use for better travel mode detection
GeoSplit: Selective Edge--Server Map Acquisition for Cross-User Travel-Mode Recognition
Machine LearningDistributed, Parallel, and Cluster Computing
Summary
Figuring out how people are traveling—like walking, biking, or driving—usually needs a lot of map information sent over the network. The authors introduce GeoSplit, a system that smartly decides when to ask for map data and when to rely on just movement info. This approach keeps accurate travel mode recognition but sends much less data, saving communication costs. They tested it with a set of user trips and found it worked well even with some users not seen in training.
What this means in practice
- •For mobile app developers: Integrate GeoSplit to improve travel mode detection accuracy while minimizing map data requests, reducing network use and app payloads in location-based services.
- •For intelligent transportation teams: Use GeoSplit to efficiently identify users’ transportation modes in real time with selective map queries, lowering backend communication overhead.
Tested on one dataset.
Authors
Zepu Wang, Luyao Niu, Ruolin Li, Jeff, Ban
Abstract
Networked mobility sensing couples local trajectory inference with remote geospatial context, making communication decisions part of the prediction system. We present GeoSplit, an edge--server framework for travel-mode recognition that separates the utility of map evidence, its training-time transfer, and its selective acquisition at runtime. On GeoLife with 10% labels, we evaluate these components under user-day and user-disjoint grouping. For unseen users, aligned full-map context raises macro-F1 from 0.7154 to 0.7864, while map-informed distillation yields a motion-only student at 0.7225. Through a compact query interface, entropy gating achieves 0.7729 macro-F1 versus 0.7738 for all queries, querying 39.5% of trajectories and reducing application payload by 60.5%. Under user-day grouping, selective and all-query scores are similarly close at a 21.6% query rate. Learned value estimation and probability fusion offer no consistent improvement. These results illustrate communication-efficient edge--server inference for an intelligent transportation sensing task.