The Toronto Intelligent Decision Engineering Laboratory
people / Lei Duan

Ph.D. Candidate
Mechanical and Industrial Engineering, University of Toronto

M.Sc. in Computing Science
Simon Fraser University, Canada


My Research: The Fusion of AI and OR

The subject I chose as the main focus of my research is an area where Artificial Intelligence (AI) and Operations Research (OR) are interconnected. That the interplay of AI and OR has enormous potential to provide support for making optimal decisions has hitherto been completely neglected. The goal of my research is by tapping into both AI and OR to develop a set of tools for the optimization of intelligent, real-time decision making process, which has many potential applications in the current service-driven economy.

Specifically, my Ph.D. thesis tackles the challenge of applying automated negotiation to solving the decision making problems encountered in areas of distributed planning, distributed scheduling, and especially, supply chain optimization. The overall efficiency of a supply chain lies in each participant's performance as well as in their coordination. Although software packages for optimization within an organization abound, only limited choices are available for supporting negotiations across organizations. My research is developing the negotiation frameworks, protocols, and strategies that will help humans to reach high-quality agreements in their negotiations either for long-term contract or short-term consensus of day-to-day operations.

This thesis has made strides to expand the project of my Master's degree, the achievement of which is a powerful, general-purpose search algorithm, called Systematic Local Search. The algorithm underwent rigorous test when it was applied to a series of challenging job shop scheduling benchmarks. Experimental results on these problem instances show that Systematic Local Search outperforms commonly-used heuristic methods such as simulated annealing and tabu search and achieves competitive performance compared to other techniques specially designed for job shop scheduling problems.

Acknowledgement

Over the past several years, I have been very fortunate to work with several brilliant researchers, especially my current supervisor, Dr. Beck. I am much indebted to their help and guidance during the time I was preparing and conducting my academic research.

If you want to share your thoughts, ideas, or visions with me, I will be very happy to get in touch with you:

Contact

Drop me a line lduanmie.utoronto.ca

Connect me to your network LinkedIn

Coming soon: See you on Google Wave
University of Toronto Mechanical and Information Engineering