Ming Shi
Research interests
-- My research strives to develop AI-powered autonomous networked systems with strong theoretical performance guarantees under practical domain-specific constraints.
-
Reinforcement learning (adversarial inputs, limited defender resources, safety and security, partial observability, human-in-the-loop, multi agent and scalability), bandit learning (power-of-2-arms, switching cost) and generative AI, with applications in Cloud-Edge AI, networking, wireless communications, robotics, autonomous driving, recommendation systems, smart healthcare, and NLP.
-
Online convex optimization and online control (switching cost, prediction, uncertainty), with applications in networking, wireless communication, cloud computing, cyber-physical systems, SDN/NFV, energy and power systems, transportation, and economics.
-
Theoretical performance analysis, e.g., competitive ratio, dynamic and static regret, sample complexity, equilibrium, convergence rate, and value of prediction.
News and Honors
Invited talk: “Autonomous Networked Systems: Adversarial Online RL Under Limited Defender Resources”
at New York University (NYU), New York, November 2023.
Invited talk: “AI-Powered Autonomous Systems with Switching Costs: Power-of-2-Arms”
at the Massachusetts Institute of Technology (MIT), Cambridge, Massachusetts, July 2023.
-
Invited talk: “Near-Optimal Adversarial Reinforcement Learning with Switching Costs”
at the California Institute of Technology (Caltech), Pasadena, California, July 2023.
-
Invited talk: “RL under Instantaneous Hard Safety Constraints and POMDPs: Robotics and Smart Health”
at Northeastern University (Northeastern), Boston, Massachusetts, July 2023.
-
Conference presentation: “A Near-Optimal Algorithm for Safe Reinforcement Learning Under Instantaneous Hard Constraints”
at International Conference on Machine Learning (ICML), Honolulu, Hawaii, July 2023.
-
Conference presentation: “Near-Optimal Adversarial Reinforcement Learning with Switching Costs”
at International Conference on Learning Representations (ICLR), Spotlight presentation, virtual, May 2023.
-
Notable-top-25% (Spotlight) paper, International Conference on Learning Representations (ICLR), January 2023.
-
Conference presentation: “Leveraging Synergies Between AI and Networking to Build Next Generation Edge Networks”
at IEEE Conference on Collaboration and Internet Computing (CIC), virtual, December 2022.
-
Conference presentation: “Power-of-2-Arms for Bandit Learning with Switching Costs”
at ACM International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (MobiHoc), Seoul, South Korea, October 2022.
-
Invited talk: “Power-of-2-Arms for Bandit Learning with Switching Costs”
at the California Institute of Technology (Caltech), Pasadena, California, May 2022.
-
Conference presentation: “Combining Regularization with Look-Ahead for Competitive Online Convex Optimization”
at IEEE Conference on Computer Communications (INFOCOM), virtual, May 2021.
-
Bilsland Dissertation Fellowship, Purdue University, April 2021.
-
IEEE INFOCOM Student Conference Award, US National Science Foundation (NSF), March 2021.
-
Conference presentation: “On the Value of Look-Ahead in Competitive Online Convex Optimization”
at ACM SIGMETRICS / IFIP Performance Joint International Conference (SIGMETRICS), Phoenix, Arizona, June 2019.
-
ACM SIGMETRICS Student Travel Grant, US National Science Foundation (NSF), May 2019.
-
Conference presentation: “Competitive Online Convex Optimization with Switching Costs and Ramp Constraints”
at IEEE Conference on Computer Communications (INFOCOM), Honolulu, Hawaii, April 2018.
-
IEEE INFOCOM Student Travel Grant, US National Science Foundation (NSF), March 2018.
|