Keynote speakers

Prof. Stefan Helber, Leibniz Universität Hannover

Title: Some thoughts on machine learning, performance evaluation and optimization

Abstract: Machine learning, as a sub-area of artificial intelligence, can be seen as the approach to distill and statistically describe underlying structures behind large data sets of well-understood observations with the overarching intention to support some type of decision-making. Since digital and connected stochastic manufacturing and service systems nowadays create such large data sets, the question arises how and to which extent machine learning can or cannot easily be used to better understand the underlying "mechanics" of those systems and to facilitate their design and operation. This requires some sort of digital twin of the considered system or process. However, while the general public may see machine learning (or artificial intelligence at large) as the final way to solve all kinds of formally untractable problems, at least my own experience is, to date, less positive and clear-cut. In the talk, I will try to shed some light on the opportunities and pitfalls of using machine learning in the context of performance evaluation and optimization, in particular regarding stochastic manufacturing and service systems.