Context-Enhanced Multi-View Trajectory Representation Learning: Bridging the Gap through Self-Supervised Models

This is a PyTorch implementation of Context-Enhanced Multi-View Trajectory Representation Learning: Bridging the Gap through Self-Supervised Models(MVTraj) for trajectory representation learning as described in our paper

Requirements

Our code is based on Python version 3.7 and PyTorch version 1.8.1. And other dependencies are given in requirement.txt.

data

We conduct our experiments on two trajectory datasets: Xi'an and Chengdu. The original data was provided by JGRM, and we further processed it by incorporating Points of Interest (POI) data to enrich its features.You can get the data here.

Code Structure

You can find our pretraining code in model_train.py, and the model architecture is defined in MVTraj.py. Before running the pretraining process, ensure that the xx.json configuration files in the config/ folder are correctly set. After pretraining, you can evaluate the model using the tasks provided in the downstream/ folder.