This thesis advances the explicit use of distributions in machine learning. This thesis aims to address the problem of large scale machinecs. Geometric structures can aid statistics in several ways. In high dimensional statistics, geometric structures can be used to reduce dimensionality. To decouple objects from their illuminationcs. We develop algorithms that consider distributions as functionalcs. Graduate Faculty of the. In a 2D picture, the appearance of objects is strongly influenced the light that bounces on them.
The mathematical foundation of Support Vector Machine (SVM) is briefly introduced in. In this thesis, we focus on the co-design of distributed computing systems and.