OR and Machine learning: research and practice

Fabrizio Detassis has happily completed his doctoral thesis on Machine Learning and Constrained Optimization. In the context of industrial problems, the application of constrained optimization is one of the most powerful, explored and exploited tools for addressing prescriptive tasks. The number of applications is huge, from logistics to transportation, packaging, manufacturing, telecommunications, planning and much more.

Optit was born with the mission of expressing the potential of Operations Research (OR) and Advanced Analytics in practical contexts, and has always kept alive over the years the relationship not only theoretical but also practical with the most important innovations of this fascinating research sector.  We are therefore very pleased to congratulate Fabrizio Detassis, who happily supported his industrial doctoral thesis, funded by Optit, at the University of Bologna, in particular with the team of Prof. Michela Milano.

The aim of the research was to study, also through practical industrial cases proposed by Optit, how Machine Learning and Constrained Optimization can be used jointly to create systems that exploit the strengths of both methods.

The interaction of these two research areas in the OR field has attracted a lot of attention in recent years, given the now considerable amount of data that has become common practice to acquire in the industrial field.

The thesis explores the most recent integration techniques of the two disciplines and introduces a new and general algorithm designed to inject knowledge into learning models by means of constraints, called “Moving Target”. The method expands existing techniques for constraint injection into a relatively simple framework, suitable for tackling very general problems.

Fabrizio continues his collaboration in Optit, helping to build new bridges between OR research and the world of Machine Learning in practical optimization applications.

Related Posts

      Verified by MonsterInsights