Willem van Jaarsveld - Deep Reinforcement Learning for Data-Driven Logistics


Companies are increasingly investing in real-time tracking of logistics resources, assets and products. To monetize this investment, they must learn to continuously make optimized decisions based on the latest available information, which requires new algorithms. We discuss how Deep Reinforcement Learning may underlie such algorithms.

Biography Willem van Jaarsveld

Willem van Jaarsveld is Associate Professor in Stochastic Optimization and Machine Learning at Eindhoven University of Technology (TU/e), in the OPAC group. His main research interest is stochastic optimization, using a diverse set of methodologies including Deep Reinforcement Learning, Stochastic Programming and Dynamic Programming. Applications areas include data-driven inventory control, supply chain management, and maintenance logistics. Apart from working towards scientific impact, he aims to make an impact in practice by working closely with several companies to facilitate the implementation of the methods he develops in practice. His research on optimization of spare parts inventories has been applied at various companies, including Fokker Services and Shell Global solutions.