Abstract:Trajectory Representation Learning (TRL) is a powerful tool for spatial-temporal data analysis and management. TRL aims to convert complicated raw trajectories into low-dimensional representation vectors, which can be applied to various downstream tasks, such as trajectory classification, clustering, and similarity computation. Existing TRL works usually treat trajectories as ordinary sequence data, while some important spatial-temporal characteristics, such as temporal regularities and travel semantics, are not fully exploited. To fill this gap, we propose a novel Self-supervised trajectory representation learning framework with TemporAl Regularities and Travel semantics, namely START. The proposed method consists of two stages. The first stage is a Trajectory Pattern-Enhanced Graph Attention Network (TPE-GAT), which converts the road network features and travel semantics into representation vectors of road segments. The second stage is a Time-Aware Trajectory Encoder (TAT-Enc), which encodes representation vectors of road segments in the same trajectory as a trajectory representation vector, meanwhile incorporating temporal regularities with the trajectory representation. Moreover, we also design two self-supervised tasks, i.e., span-masked trajectory recovery and trajectory contrastive learning, to introduce spatial-temporal characteristics of trajectories into the training process of our START framework. The effectiveness of the proposed method is verified by extensive experiments on two large-scale real-world datasets for three downstream tasks. The experiments also demonstrate that our method can be transferred across different cities to adapt heterogeneous trajectory datasets.

Transfer-learning-based representation learning for trajectory similarity search

Towards robust trajectory similarity computation: Representation-based spatio-temporal similarity quantification

Representation Learning With Multi-Level Attention for Activity Trajectory Similarity Computation

Spatial-temporal fusion graph framework for trajectory similarity computation

Spatio-Temporal Trajectory Similarity Measures: A Comprehensive Survey and Quantitative Study

Learning Universal Trajectory Representation via a Siamese Geography-Aware Transformer

Deep Representation Learning of Activity Trajectory Similarity Computation.

Spatio-Temporal Trajectory Similarity Learning in Road Networks

CSTRM: Contrastive Self-Supervised Trajectory Representation Model for trajectory similarity computation

Trajectory Similarity Learning with Auxiliary Supervision and Optimal Matching

Self-supervised contrastive representation learning for large-scale trajectories

Revisiting CNNs for Trajectory Similarity Learning

Deep Spatially and Temporally Aware Similarity Computation for Road Network Constrained Trajectories

Deep Trajectory Similarity Model: A Fast Method For Trajectory Similarity Computation

Trajectory-as-a-Sequence: A Novel Travel Mode Identification Framework

A Deep Trajectory Clustering Method Based on Sequence‐to‐sequence Autoencoder Model

GRLSTM: Trajectory Similarity Computation with Graph-Based Residual LSTM

Self-supervised Trajectory Representation Learning with Temporal Regularities and Travel Semantics

Jointly spatial-temporal representation learning for individual trajectories

Convolutional Trajectory Similarity Model: a Faster Method for Trajectory Similarity Measurement.