Representation Learning with Multi-level Attention for Activity Trajectory Similarity Computation
KAUST DepartmentComputer Science Program
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Machine Intelligence & kNowledge Engineering Lab
Permanent link to this recordhttp://hdl.handle.net/10754/664294
MetadataShow full item record
AbstractMassive trajectory data stem from the prevalence of equipment-supporting GPS and wireless communication technology. Especially, activity trajectory from LBSN endows traditional trajectory data with additional user semantic activities. Measuring the similarity between activity trajectories is to compare their proximity in multiple dimensions such as time, location, and semantics. In this way, we can mine implicit user preference and apply it to route planning, POI recommendation or any other online tasks. The key challenge of comparing activity trajectories lies in two aspects. One is the uneven sampling rate in both time and space. The other is the discrepancy of individual activities. Previous effort alleviates the issue of uneven sampling rate via trajectory complements, which is limited to spatial-temporal information. In this paper, we propose to learn a representation for one activity trajectory by jointly considering the spatio-temporal characteristics and the activity semantics. The similarity of two trajectories is computed by weighting individual trajectory points and contextual features with multi-level attention mechanisms. In specific, we propose a point-level and feature-level attention mechanism to adaptively select critical elements and contextual factors for learning trajectory representation. Our proposed approach, called At2vec, demonstrates better performance than existing baselines in extensive experimental evaluation on real databases.
CitationLiu, A., Zhang, Y., Zhang, X., Liu, G., Zhang, Y., Li, Z., … Zhou, X. (2020). Representation Learning with Multi-level Attention for Activity Trajectory Similarity Computation. IEEE Transactions on Knowledge and Data Engineering, 1–1. doi:10.1109/tkde.2020.3010022