EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Robustness of localized learning Authors:  Andreas Christmann - University of Bayreuth (Germany) [presenting]
Abstract: The computation of kernel methods is usually not fast enough for big data sets. A localized learning method based on kernels is investigated. Recent results on universal consistency and on statistical robustness properties of such localized learning methods will be given.