Optimization and Machine Learning: a two way trip with applications
Optimization methods are a key element when training Machine Learning models. Typically, certain refinements of stochastic gradient descent methods are mostly used, to find reasonably good configurations that minimize the given loss functions. Conversely, Machine Learning methods are also recently used to tackle paradigmatic problems in optimization, mostly combinatorial optimization problems. OPTIMALE project aims to explore these connections. As a by product, some applications with companies in the region such as Grupo Energético Puerto Real S.A. and Eléctrica de Puerto Real S.A. involve prediction of consumption, water leakage and non-intrusive load monitoring with energy and water data provided by these companies.