The parameter settings of different ablation experiments refer to the run.py file
python run.py --data_type tdrive --remask 0 --d_model 512 --if_posiemb 1 --loss spatial_loss --use_his 1 --is_training 1 --gpu 0 --epoch 50 --lr 1e-4python run.py --data_type geolife --remask 2 --d_model 512 --if_posiemb 1 --loss spatial_loss --use_his 1 --is_training 1 --gpu 0 --epoch 50 --lr 1e-4python run.py --data_type cdr --remask 0 --d_model 512 --if_posiemb 1 --loss spatial_loss --use_his 1 --is_training 1 --gpu 0 --epoch 50 --lr 1e-4If you find this repo useful and would like to cite it, citing our paper as the following will be really appropriate:
@ARTICLE{10189092,
author={Si, Junjun and Yang, Jin and Xiang, Yang and Wang, Hanqiu and Li, Li and Zhang, Rongqing and Tu, Bo and Chen, Xiangqun},
journal={IEEE Transactions on Mobile Computing},
title={TrajBERT: BERT-Based Trajectory Recovery with Spatial-Temporal Refinement for Implicit Sparse Trajectories},
year={2023},
volume={},
number={},
pages={1-12},
doi={10.1109/TMC.2023.3297115}}