88304am永利集团

Neural combinatorial optimization (NCO) has emerged as a promising paradigm for solving combinatorial optimization problems by leveraging the power of deep learning, which offers a data-driven, and adaptive approach that learns from problem instances and demonstrates strong generalization capabilities for effectively handling a wide range of problem instances. This talk focuses on the applications of NCO in solving practical vehicle routing problems (VRPs), which are fundamental in logistics and transportation. Several representative NCO models for VRPs will be reviewed firstly. I will then highlight how we developed NCO models to address VRPs with complex real-world constraints, such as time-dependent travel speeds and uncertain parking availability, which are challenging for traditional techniques. The results of performance comparisons between our NCO models and several benchmark models will be presented. Finally, significant challenges in applying NCO to practical VRPs will be discussed.

郭钊侠，香港理工大学博士，现任四川大学商学院教授、博士生导师，工业工程与管理系系主任。主要从事大数据与人工智能算法、复杂系统建模与优化等相关交叉领域的研究。入选四川省学术和技术带头人、四川省数字政府项目专家组专家。

主编出版人工智能算法教材1部、专著3部，发表论文100余篇，代表性成果发表在Cell姊妹刊One Earth、The Lancet Planetary Health、Nature Communications、INFORMS Journal on Computing等期刊。成果曾获细胞出版社2023中国年度论文、2024“数据要素×”大赛全国总决赛三等奖（城市治理赛道）、2017年度四川省科技进步二等奖（第1完成人）等科技奖项，并被央视等媒体报道以及联合国等国际组织报告引用与采纳。