Selective inference is a problem that can be spotted in several machine learning tasks. Although some solutions to this problem have already been proposed, these solutions consider the use of algorithms which have been developed to solve more general problems of inference. However, recent work in machine learning shows that it is important to solve a specific inference problem, such as selective inference, directly, without relying on solutions to more general problems. To this end, this work proposes researching and developing algorithms to specifically solve the selective inference problem.
News published in Agência FAPESP Newsletter about the scholarship: