This thesis explores and expands the use of spatially adaptive sparse grids for solving various learning tasks, in a multi-faceted approach. We introduce new estimators of functions of probability densities, we improve time series prediction results on financial data by means of a regression/classification code optimized for a specific computer architecture, and we introduce two new clustering techniques, implemented in a user-oriented manner, which can analyze deterministic and uncertain data, respectively.
«
This thesis explores and expands the use of spatially adaptive sparse grids for solving various learning tasks, in a multi-faceted approach. We introduce new estimators of functions of probability densities, we improve time series prediction results on financial data by means of a regression/classification code optimized for a specific computer architecture, and we introduce two new clustering techniques, implemented in a user-oriented manner, which can analyze deterministic and uncertain data,...
»