add krr page in cookbook #3078
add krr page in cookbook #3078
Conversation
| Kernel Ridge Regression | ||
| ======================= | ||
|
|
||
| Kernel ridge regression is a kernel-based regularized form of regression which learns a function in the space induced by the respective kernel and the data by minimizing a squared error loss with :math:`L_2` regularization. For linear ridge regression, it boils down to solving a linear system: |
There was a problem hiding this comment.
Kernel ridge regression is a non-parametric form of ridge regression. The aim is to learn a function in the space induced by the respective kernel
|
I think there also will be a description for krr, which can be removed |
|
@karlnapf I don't fully understand your last comment. |
|
Check the descriptions folder in the examples dir On Tuesday, 15 March 2016, Sanuj Sharma notifications@github.com wrote:
|
|
@karlnapf removed the description file. |
|
Cool this is good to merge. But I think it is good to explain such things too. |
| {\bf \alpha} = \left({\bf K}+\tau{\bf I}\right)^{-1}{\bf y} | ||
|
|
||
| where :math:`{\bf K}` is the kernel matrix and :math:`{\bf \alpha}` is the vector of weights in the space induced by the kernel. | ||
| The learned function can then be evaluated as :math:`f(x)=\sum_{i=1}^Nk(x,x_i)`. |
There was a problem hiding this comment.
the alpha_i are missing here .. sorry I forgot above
There was a problem hiding this comment.
yeah, i was thinking the same but was not sure. Updating.
|
@karlnapf Updated |
@karlnapf looking forward to your comments :)