Publication Type

Conference Proceeding Article

Publication Date



Modeling trajectory data is a building block for many smart-mobility initiatives. Existing approaches apply shallow models such as Markov chain and inverse reinforcement learning to model trajectories, which cannot capture the long-term dependencies. On the other hand, deep models such as Recurrent NeuralNetwork (RNN) have demonstrated their strength of modeling variable length sequences. However, directly adopting RNN to model trajectories is not appropriate because of the unique topological constraints faced by trajectories. Motivated by these findings, we design two RNN-based models which can make full advantage of the strength of RNN to capture variable length sequence and meanwhile to address the constraints of topo-logical structure on trajectory modeling. Our experimental study based on real taxi trajectory datasets shows that both of our approaches largely outperform the existing approaches.


Artificial intelligence, Inverse problems, Markov processes, Reinforcement learning, Taxicabs, Topology, Trajectories, Building blockes, Inverse reinforcement learning, Logical structure, Long-term dependencies, Recurrent neural network (RNN), Topological constraints, Trajectory data, Trajectory modeling, Recurrent neural networks


Computer Engineering | Databases and Information Systems | Transportation

Research Areas

Data Management and Analytics


Proceedings of the 26th International Joint Conference on Artificial Intelligence IJCAI-17, Melbourne, Australia, August 19-25

First Page


Last Page




City or Country

Los Altos, CA

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.