Giulia’s goal is to use Machine Learning techniques to improve the way we do optimization, in particular with Mixed Integer Programs. She has started her research with a classification question: can we classify different types of Quadratic MIP instances depending on some problem’s features? The aim is to consequently solve them with the most appropriate algorithm, and possibly integrate the Machine Learning process with the optimization solvers’ one.
In general, her research will be focus on those optimization situations in which we need to take a heuristic decision and we could potentially learn the decision itself using Machine Learning
A. Lodi, G. Zarpellon. 2017. On learning and branching: a survey. Accepted in the TOP Journal. DS4DM-2017-004