Buổi 2: Thứ 5 ngày 7/4/2022 (19h30 - 21h00)

Talk 3: Evolutionary Learning for Combinatorial Optimisation

Abstract:

Combinatorial optimisation is ubiquitous in Artificial Intelligence and Operations Research. Many important real-world applications from timetabling, production scheduling, fleet management, to mission planning can be formulated as combinatorial optimisation problems. The goal of solving these problems is to help decision makers find solutions that optimise an objective of interest such as productivity, resource utilisation, and costs. However, there is no scalable solver or algorithm that can universally solve these problems. Research in combinatorial optimisation usually “searches” for “tricks” or “heuristics” to enhance the efficiency of optimisation algorithms, which is a very time-consuming iterative process, and the generalisation of these techniques is often questionable. Machine learning has been recently examined as a way to automate the “search” for such “tricks” in combinatorial optimisation. This talk focuses on evolutionary learning, a branch in AI inspired by natural evolution, and its applications to combinatorial optimisation. Applications of the evolutionary learning approach in production scheduling and logistics optimisation will be presented. This talk also discusses the technical challenges and opportunities to combine evolutionary learning, deep learning, and optimisation solvers to create an innovative optimisation framework with self-adaptive abilities and human-in-the-loop mechanisms.

Bio:

Su Nguyen is a Senior Lecturer in Business Analytics and Artificial Intelligence at the Centre for Data Analytics and Cognition (CDAC), La Trobe University, Australia. He received his Ph.D. degree in Artificial Intelligence and Operations Research from Victoria University of Wellington (VUW), Wellington, New Zealand, in 2013. His expertise includes simulation-optimization, evolutionary computation, automated algorithm design, interfaces of artificial intelligence and operations research, and their applications in logistics, energy, and transportation. Nguyen has a strong track record in developing simulation models, simulation-based decision support tools, and simulation-optimisation algorithms for industry applications. He has 70+ publications in top peer-reviewed journals and conferences in computational intelligence and operations research. His current research focuses on hybrid intelligence systems that combine the power of modern artificial intelligence technologies and operations research methodologies. He was the chair (2014-2018) of IEEE task force on Evolutionary Scheduling and Combinatorial Optimisation and is a member of IEEE CIS Data Mining and Big Data technical committee. He delivered tutorials about evolutionary simulation-optimisation and AI-based visualisation at Parallel Problem Solving from Nature Conference (2018) and IEEE World Congress on Computational Intelligence (2020).