| Popis: |
This paper is an announcement for our longer paper in preparation. Traditional kernel based methods utilize either a fixed kernel or a combination of judiciously chosen kernels from a fixed dictionary. In contrast, we construct a data-dependent kernel utilizing the components of the eigen-decompositions of different kernels constructed using ideas from diffusion geometry, and use a regularization technique with this kernel with adaptively chosen parameters. In this paper, we illustrate our method using the two moons dataset, where we obtain a zero test error using only a minimal number of training samples. |
