This Ph.D., falling under the ENW- Groot project OPTIMAL (Optimization for and with Machine Learning), aims to overcome some disadvantages of classical optimization models through the use of machine learning.
The currently used manual model building process requires experts in optimization, needs constant adaptation to changing situations, is not data-driven and sometimes can be very difficult to realize.
The goal is to start from the data and use machine learning to define part of the optimization process, making it easier to develop and understand. Hence, this results in Data-centric optimization models that contain, e.g., deep learning, random forest, support vector machines, or symbolic regression models. There are several key aspects to investigate, from data collection to the definition of which machine learning model is the most appropriate or how to integrate it into optimization models.
The two main applications are related to the World Food Programme project and Cancer treatment optimization.