👋 About Me

I am a Ph.D. student at the Southern University of Science and Technology (SUSTech), supervised by Zhenkun Wang and Xi Lin. My research focuses on AI for Optimization, specifically Learning to Optimize, also known as Neural Combinatorial Optimization. This paradigm leverages neural networks to solve complex combinatorial optimization problems (COPs) such as routing, scheduling, and assignment. By learning policies from data rather than relying on manual design, we aim to develop fast solvers that generalize across problem scales, distributions, and diverse problems. It is a highly active and fascinating field at the intersection of AI and Operations Research, presenting both great potential and exciting challenges.

My research focuses on deep reinforcement learning for vehicle routing problems (VRPs), with an emphasis on out-of-distribution zero-shot generalization and domain foundation model construction. My recent first/co-first-author work includes the following: (1) ICAM, an instance-conditioned adaptation model for generalization to thousand-scale VRP instances; (2) L2R, a learning-based search space reduction framework scalable to 10-million-node instances; (3) URS, a powerful domain foundation model capable of solving 100+ VRP variants; and (4) SPACE, a generalist neural solver that unifies symmetric and asymmetric routing problems through coordinate-independent representations.

As of September 2026, I have 9 accepted papers, including 5 first/co-first-author papers. My work has appeared in venues including NeurIPS, ICML, KDD, and IEEE T-ITS.

🔥 News

• 🎉 09/2026: SPACE was accepted to NeurIPS 2026.
• 🎉 05/2026: L2R appeared at the 32nd SIGKDD Conference on Knowledge Discovery and Data Mining (KDD).
• 🎉 05/2026: URS appeared at the 43rd International Conference on Machine Learning (ICML).
• 🎉 03/2026: ICAM appeared in IEEE Transactions on Intelligent Transportation Systems (T-ITS).
• 🎓 09/2022: Began Ph.D. study at Southern University of Science and Technology.

🎯 Research Highlights

ICAM: Instance-Conditioned Adaptation Model for Thousand-Scale Generalization

T-ITS 2026

📄 Paper 💻 Code

• Introduced a lightweight instance-conditioned adaptation function and a low-complexity attention mechanism to adjust policies using instance-specific information.
• Validated the method on four representative routing problems (TSP, CVRP, ATSP \& CVRPTW), improving generalization while preserving low inference time (at the second level).

L2R: Learning to Reduce Search Space for Million-Scale Generalization

KDD 2026

📄 Paper 💻 Code

• Proposed the first learning-based framework to dynamically prune the search space by adaptively prioritizing nodes based on problem-specific features and states.
• Developed the first specialist neural model for 10-million-node VRP instances. Trained only on 100-node instances, it generalizes to TSP10M in a zero-shot manner with a 5.05% gap.

URS: A Powerful Domain Foundation Model for Cross-Problem Zero-Shot Generalization

ICML 2026

📄 Paper 💻 Code

• Proposed a unified data representation that eliminates the need for problem enumeration by projecting diverse VRP variants into a unified feature space.
• Developed the first domain foundation model that efficiently solves over 100 VRP variants with a single model without retraining or fine-tuning, while scaling to 7,000-node instances.

SPACE: Unifying Symmetric and Asymmetric Routing Problems for Generalist Neural Solver

NeurIPS 2026

📄 Paper

• Combined bidirectional Fréchet representations and weight-decomposed adaptive decoding for zero-shot generalization.
• Achieved coordinate-independent unified feature encoding, improving asymmetric routing while remaining competitive on symmetric problems.

MiLoop: Selective Memory Propagation for Neural Combinatorial Optimization

arXiv 2026

📄 Paper

• Selective memory propagation reuses and updates node representations across RL construction steps, enabling a shallow policy without external solution labels.
• Evaluated on four combinatorial optimization problems, with tested instance sizes ranging from 100 to 10 million nodes.

📝 Publications

Arxiv Preprint

1. Zhuoliang Xie, Changliang Zhou, Genghui Li, Zhenkun Wang^†. Multi-Task Evolution for Zero-Shot Cross-Problem Generalization using LLMs. arXiv preprint arXiv:2610.03316, 2026.

   📄 Paper

2. Changliang Zhou*, Yuanyao Chen*, Rongsheng Chen, Zhiyun Lin, Zhenkun Wang^†. MiLoop: Selective Memory Propagation for Neural Combinatorial Optimization. arXiv preprint arXiv:2610.01685, 2026.

   📄 Paper

3. Canhong Yu, Changliang Zhou, Rongsheng Chen, Zhenkun Wang^†, Yu Zhou. Rethinking Constraint Awareness for Efficient State Embedding of Neural Routing Solver. arXiv preprint arXiv:2605.10122, 2026.

   📄 Paper

4. Yunpeng Ba, Xi Lin, Changliang Zhou, Ruihao Zheng, Zhenkun Wang^†, Xinyan Liang, Zhichao Lu, Jianyong Sun, Yuhua Qian, Qingfu Zhang. Survey on Neural Routing Solvers. arXiv preprint arXiv:2602.21761, 2026.

   📄 Paper

Accepted Conference Papers:

1. NeurIPS 2026 Rongsheng Chen*, Changliang Zhou*, Canhong Yu, Yuanyao Chen, Yu Zhou, Zhuo Chen, Zhenkun Wang^†. SPACE: Unifying Symmetric and Asymmetric Routing Problems for Generalist Neural Solver. Advances in Neural Information Processing Systems (NeurIPS), 2026.

   📄 Paper

2. ICML 2026 Changliang Zhou, Canhong Yu , Shunyu Yao, Xi Lin, Zhenkun Wang^†, Yu Zhou, Qingfu Zhang. URS: A Unified Neural Routing Solver for Cross-Problem Zero-Shot Generalization. 43rd International Conference on Machine Learning (ICML), 2026.

   📄 Paper 💻 Code

3. KDD 2026 Changliang Zhou*, Xi Lin*, Zhenkun Wang^†, Qingfu Zhang. Learning to Reduce Search Space for Generalizable Neural Routing Solver. 32nd SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2026.

   📄 Paper 💻 Code

4. NeurIPS 2024 Zhi Zheng, Changliang Zhou, Xialiang Tong, Mingxuan Yuan, Zhenkun Wang^†. UDC: A Unified Neural Divide-and-Conquer Framework for Large-Scale Combinatorial Optimization Problems. Advances in Neural Information Processing Systems (NeurIPS), 2024.

   📄 Paper 💻 Code

5. AAAI 2024 Yubin Xiao, Di Wang, Boyang Li, Mingzhao Wang, Xuan Wu, Changliang Zhou, You Zhou^†. Distilling Autoregressive Models to Obtain High-Performance Non-autoregressive Solvers for Vehicle Routing Problems with Faster Inference Speed. Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2024.

   📄 Paper 💻 Code

Accepted Journal Articles:

1. T-ITS 2026 Changliang Zhou*, Xi Lin*, Zhenkun Wang^†, Xialiang Tong, Mingxuan Yuan, Qingfu Zhang. Instance-Conditioned Adaptation for Large-Scale Generalization of Neural Routing Solver. IEEE Transactions on Intelligent Transportation Systems (T-ITS), 2026.

   📄 Paper 💻 Code

2. JWPE 2025 Sheng Miao, Xuefei Li, Huaying Sun, Xiubo Chen, Changliang Zhou, Xiang Shen, Chao Liu^†, Changqing Liu, Weijun Gao. Multi-output Behavioral Cloning Framework: A Knowledge-Based Predictive Control Methodology Based on Deep Learning for Wastewater Treatment Plants. Journal of Water Process Engineering (JWPE), 69:106813, 2025.

   📄 Paper

3. SCS 2021 Sheng Miao*^†, Changliang Zhou*, Salman Ali AlQahtani, Mubarak Alrashoud, Ahmed Ghoneim, Zhihan Lv. Applying Machine Learning in Intelligent Sewage Treatment: A Case Study of Chemical Plant in Sustainable Cities. Sustainable Cities and Society (SCS), 72:103009, 2021.

   📄 Paper

4. JCR 2020 Chao Liu^†, Haoqing Zhang, Sheng Miao, Xuefei Li, Qun Miao, Changliang Zhou. Assessment of Status and Vulnerability to Seawater Intrusion in Wendeng District, China. Journal of Coastal Research (JCR), 104(sp1):546-553, 2020.

   📄 Paper

* Equal contribution. ^† Corresponding author. xxxx names indicate co-mentored students.

(Last updated October 2026.)

🛠 Open Source Projects

🎓 Education

• Sep. 2022 - Present, Southern University of Science and Technology, Ph.D. student in Intelligent Manufacturing and Robotics, School of Automation and Intelligent Manufacturing.
• Sep. 2019 - Jun. 2022, Qingdao University, M.Eng. degree in Software Engineering, College of Computer Science and Technology.
• Sep. 2014 - Jun. 2018, Linyi University, B.Eng. degree in Software Engineering, School of Information Science and Engineering.

🏆 Service and Honors

• Reviewer: ICLR 2026, 2027; NeurIPS 2026 (Top Reviewer ); IEEE T-ITS; TMLR; IEEE TNNLS.
• University Honor: Outstanding Postgraduate Student, Southern University of Science and Technology, 2022-2023.

👨‍🏫 Teaching

• Teaching Assistant, SDM374 Machine Learning System Design, Southern University of Science and Technology (Fall 2023), Sep. 2023 - Jan. 2024.
• Teaching Assistant, SDM374 Machine Learning System Design, Southern University of Science and Technology (Fall 2022), Sep. 2022 - Jan. 2023.