About
I will join Nanjing University as an incoming PhD student in Fall 2026, advised by Prof. Guanghui Zhu. Prior to this, I obtained an M.Sc. in Software Engineering under the supervision of Prof. Yafei Li, and a B.Eng. in Mechanical Engineering from Zhengzhou University.
My current research focuses on the training and fine-tuning of large language models, including learning, adaptation, reasoning, and generalization. I am also interested in reinforcement learning and preference-based learning methods, particularly for improving alignment, robustness, and adaptation in complex decision-making systems. I am open to research collaborations related to these areas of interest.
Selected Publications
View All βCredit Assignment and Fine-Tuning Enhanced Reinforcement Learning for Collaborative Spatial Crowdsourcing
Wei Chen, Yafei Li, Baolong Mei, Guanglei Zhu, Jiaqi Wu, Mingliang Xu
Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence
We propose CAFE, a multi-agent RL framework for spatial crowdsourcing that addresses delayed rewards and non-stationary distributions through credit assignment mechanisms and adaptive fine-tuning, achieving superior task completion and equitable reward distribution.
Gradient-Guided Credit Assignment and Joint Optimization for Dependency-Aware Spatial Crowdsourcing
Yafei Li, Wei Chen, Jinxing Yan, Huiling Li, Lei Gao, Mingliang Xu
Proceedings of the AAAI Conference on Artificial Intelligence
We propose RMO, a two-stage framework for dependency-aware spatial crowdsourcing that uses multi-agent RL for subtask recommendation and utility-based matching, employing meta-gradients and gradient synchronization to address credit assignment and joint optimization challenges.
Effective Task Assignment in Mobility Prediction-Aware Spatial Crowdsourcing
Huiling Li, Yafei Li, Wei Chen, Shuo He, Mingliang Xu, Jianliang Xu
2025 IEEE 41st International Conference on Data Engineering (ICDE)
We address Task Assignment in Mobility Prediction-aware Spatial Crowdsourcing (TAMP) through a task-adaptive meta-learning algorithm that clusters workers and trains mobility prediction models, coupled with a task assignment algorithm that prioritizes high-confidence completions, achieving improved assignment quality.
Multi-aircraft cooperative decision-making methods driven by differentiated support demands for carrier-based aircraft
Wei Chen, Lulu Li, Dong Chen, Yafei Li, Ke Wang, Yuanyuan Jin, Mingliang Xu
ACTA AERONAUTICAET ASTRONAUTICA SINICA
We propose DATSDM, a novel Dependency-Aware Task Scheduling Decision Module leveraging graph neural networks and Transformer attention mechanisms
News
Our paper Credit Assignment and Fine-Tuning Enhanced Reinforcement Learning for Collaborative Spatial Crowdsourcing has been accepted at IJCAI 2025 π.
Our paper Gradient-Guided Credit Assignment and Joint Optimization for Dependency-Aware Spatial Crowdsourcing has been accepted as an oral presentation at AAAI 2025 π .