Speaker: Meng Xu, A*STAR - Agency for Science, Technology and Research, Singapore
Date: 17 February 2025
Time: 4:00 - 5:00pm (Beijing Time, UTC+8)
Dr. Meng Xu received her B.S. and M.S. degrees from Beijing Institute of Technology, Beijing, China, in 2017 and 2020, respectively, and her Ph.D. degree from the Centre for Data Science and Artificial Intelligence & School of Engineering and Computer Science, Victoria University of Wellington, New Zealand, in 2024. Since July 2024, she has been a Scientist at the Singapore Institute of Manufacturing Technology (SIMTech), part of the Agency for Science, Technology and Research (A*STAR), Singapore. Concurrently, she serves as a Postdoctoral Fellow at the College of Computing and Data Science, Nanyang Technological University, Singapore. Dr. Xu's research interests include evolutionary computation, genetic programming, scheduling, reinforcement learning, and evolutionary multi-objective optimization. She has published in top-tier journals and international conferences, including IEEE TEVC, IEEE TSC, and IEEE CIM. For more information, visit her website at https://mengxu95.github.io/.
Dynamic flexible job shop scheduling (DFJSS) has garnered significant attention from both academia and industry due to its extensive industrial applications and real-world impact. The core challenge in DFJSS lies in simultaneously addressing machine assignment and operation sequencing in dynamic environments, where conditions such as the arrival of new jobs evolve over time. Among existing methods, scheduling heuristics have emerged as effective solutions, widely adopted for their simplicity and real-time responsiveness. However, designing scheduling heuristics is both time-intensive and reliant on substantial domain knowledge. Genetic Programming (GP), as a hyper-heuristic approach, has shown considerable success in automatically evolving scheduling heuristics for DFJSS. Despite these achievements, there remains potential for further advancements to enhance GP's performance in this domain. This talk highlights our recent progress in applying GP to DFJSS, with a focus on: diversity-based parent selection mechanisms, joint decision-making mechanisms via ensemble learning, collaborative heuristic generation and selection by integrating GP with reinforcement learning, and multi-objective problem-solving capabilities.
The video recording of the Webinar can be found here.
The slides of the Webinar can be found here.
Back to the Webinar Series