Spatio-temporal feature fusion for dynamic taxi route recommendation via deep reinforcement learning

作者：

Highlights：

• We comprehensively study the spatio-temporal features for taxi route recommendation.

• A deep policy network is carefully designed to fuse the extracted features.

• An adaptive deep reinforcement learning method is developed to learn the policy net.

• Evaluation using real-world datasets demonstrates the effectiveness of our method.

摘要

•We comprehensively study the spatio-temporal features for taxi route recommendation.•A deep policy network is carefully designed to fuse the extracted features.•An adaptive deep reinforcement learning method is developed to learn the policy net.•Evaluation using real-world datasets demonstrates the effectiveness of our method.

论文关键词：Spatio-temporal feature fusion,Sequential decision making,Taxi route recommendation,Deep reinforcement learning,Transportation

论文评审过程：Received 29 February 2020, Revised 6 July 2020, Accepted 20 July 2020, Available online 25 July 2020, Version of Record 27 July 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106302