613 search results - page 3 / 123 » The Pre-Image Problem in Kernel Methods |
JMLR
2010
13 years 2 days ago
2010
The problem of automatic feature selection/weighting in kernel methods is examined. We work on a formulation that optimizes both the weights of features and the parameters of the ...
ICDM
2010
IEEE
13 years 3 months ago
2010
IEEE
Financial forecasting is the basis for budgeting activities and estimating future financing needs. Applying machine learning and data mining models to financial forecasting is both...
PR
2010
13 years 3 months ago
2010
Most existing representative works in semi-supervised clustering do not sufficiently solve the violation problem of pairwise constraints. On the other hand, traditional kernel met...
NECO
1998
13 years 4 months ago
1998
We describe a new method for performing a nonlinear form of Principal Component Analysis. By the use of integral operator kernel functions, we can e ciently compute principal comp...
PAMI
2011
13 years 6 days ago
2011
— Kernel mapping is one of the most used approaches to intrinsically derive nonlinear classifiers. The idea is to use a kernel function which maps the original nonlinearly separ...