Marta Fochesato

Marta Fochesato is a Ph.D. candidate at the Automatic Control Laboratory, ETH Zürich, under Prof. Lygeros. Her research interests lie at the intersection of data-driven optimization under uncertainty, control, and energy-related applications.

She works on distributional robust control to robustify sequential decision-making processes under uncertainty against input misspecifications in the probability distribution. To this end, she constructs a set of probability distributions that are “likely” to have generated the observed data, and designs a control policy that performs “well” across all possible distributions in this set. Via a principled design of this set, this framework accounts for statistical errors and distribution shifts, for example, temperature shifts due to climate change or endogenous shifts due to our actions on the system. She is applying these methods primarily to energy systems and markets, such as the day-ahead bidding problem.