Embedded Files
Scalable Learning-Based Control and Optimization
1. Zeroth-Order Methods for Model-Free Control and Optimization
Zeroth-order (ZO) methods are prominent approaches for solving black-box optimization and control problems using only function evaluations (zeroth-order information) or system feedback. We aim to accelerate the convergence of ZO methods and advance them to constrained and distributed settings, enabling scalable applications for the optimization and control of complex multi-agent cyber-physical systems.
Featured Publications:
[1] Xin Chen, Zhaolin Ren, “Regression-Based Single-Point Zeroth-Order Optimization“, arXiv:2507.04223, 2025.
[2] Xin Chen, Jorge I. Poveda, Na Li, “Continuous-Time Zeroth-Order Dynamics with Projection Maps: Model-Free Feedback Optimization with Safety Guarantees”, IEEE Transactions on Automatic Control, vol. 70, no. 8, pp. 5005-5020, Aug. 2025.
[3] Xin Chen, Yujie Tang, Na Li, “Improve Single-Point Zeroth-Order Optimization Using High-Pass and Low-Pass Filters”, 39th International Conference on Machine Learning (ICML), Baltimore, MD, USA, 2022.
[4] Xin Chen, Jorge I. Poveda, Na Li, “Safe Model-Free Optimal Voltage Control via Continuous-Time Zeroth-Order Methods”, 60th IEEE Conference on Decision and Control (CDC), Austin, Texas, USA, 2021. (Outstanding Student Paper Award) [Full Version]
2. Reinforcement Learning and Advanced Multi-Armed Bandits
Reinforcement learning (RL) is a prominent machine learning paradigm concerned with how agents take sequential actions in an uncertain interactive environment and learn from the feedback to optimize a specific performance. We aim to explore RL applications in physical energy systems and understand the advantages and limitations. To enhance scalability, we study a simplified version of RL, Multi-Armed Bandits (MABs), and advance the MAB framework to contextual and restless settings that incorporate external environmental influence and internal system dynamics.
Featured Publications:
[1] Xin Chen, I-hong Hou, “Contextual Restless Multi-Armed Bandits with Application to Demand Response Decision-Making“, IEEE Conference on Decision and Control (CDC), Milan, Italy, 2024.
[2] Xin Chen, Guannan Qu, Yujie Tang, Steven Low, Na Li, “Reinforcement Learning for Selective Key Applications in Power Systems: Recent Advances and Future Challenges”, IEEE Transactions on Smart Grid, vol. 13, no. 4, pp. 2935-2958, July 2022. (IEEE Transactions on Smart Grid Top-5 Outstanding Paper in 2020 – 2022)
[3] Xin Chen, Yingying Li, Jun Shimada, Na Li, “Online Learning and Distributed Control for Residential Demand Response”, IEEE Transactions on Smart Grid, vol. 12, no. 6, pp. 4843-4853, Nov. 2021.
[4] Xin Chen, Yutong Nie, Na Li, “Online Residential Demand Response via Contextual Multi-Armed Bandits,” IEEE Control Systems Letters, vol. 5, no. 2, pp. 433-438, April 2021.
Page updated
Google Sites
Report abuse