Most real-world problems, such as those found in production systems engineering, are inherently multi-criteria by nature so that multiple optimization objectives must be handled effectively. Multi-objective optimization (MOO) algorithms, which can generate multiple optimal trade-off solutions for the decision-maker to choose from, have been increasingly used in practice during the last two decades.
Despite the increasing popularity of applying MOO in many different domains, it is less discussed the connection between MOO and knowledge discovery. Since a MOO process involves the generation of a trade-off optimal solution set, instead of a single optimal solution, the solution set can then be used (1) for humans to learn the relationships among the conflicting objectives, between the decision variables and the objectives, as well as any interactions among the decision variables; (2) to be fed as training samples for machine learning algorithms to discover any knowledge, in the form of rules, salient and/or hidden patterns, about what constitute the optimal trade-off solutions.
Apart from supporting the decision-maker to gain a better understanding of the problem and make confident decisions, the extracted knowledge can also be stored for future use in related optimization and decision-making scenarios. Such a process of extracting, learning, storing, and re-using the knowledge enabled by MOO is generally referred to as Knowledge-Driven Optimization (KDO). In this talk, we will introduce KDO with a particular focus on its applications in production systems design, analysis, and improvement. We will also discuss the covered topics from a human and machine co-learning perspective.
This event is being hosted by the NAFEMS Optimisation Working Group (OWG). The OWG has formed an online Community to help disseminate best practice and encourage the adoption of optimisation methods and technology. For more information and to get involved go to the Optimisation Community webpage.
This webinar is available for free to the engineering analysis community, as part of NAFEMS' efforts to bring the community together online.
Professor Amos Ng, University of Skövde, Sweden.
Amos H.C. Ng is a professor in Production and Automation Engineering at the University of Skövde, Sweden. He is also a visiting professor in the Division of Industrial Engineering and Management at Uppsala University, Sweden, and the CEO of Evoma AB.
Since 2005, Amos has been conducting research and teaching in Factory Physics, production simulation, and simulation-based optimization, particularly the applications of multi-objective simulation optimization for production systems design, analysis, and improvement. The technology and software he developed with his team have earned him a Volvo Car Technology Award in 2013 and two best conference paper awards (Sweden and Italy) in 2012-13.
Amos' current research focus lies in combining multi-disciplinary simulations and AI-based prescriptive analytics for supporting decision-making in manufacturing/supply-chain/health-care domains. In conjunction with this is his leading role as the principal investigator of the profile project called Virtual Factories and Knowledge-Driven Optimization, which started in 2018 and will last till 2026, with the collaborations of many major manufacturing firms in Sweden.