Skip to content

ICML paper: Robust and Efficient Kernel Hyperparameter Paths with Guarantees

License

Notifications You must be signed in to change notification settings

PatWie/RobustKernelPaths

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Robust and Efficient Kernel Hyperparameter Paths with Guarantees

**auto finetuning, svm, parameters, kernel parameter, choose parameters **

The results of non-linear SVM classifications heavily rely on the choose of the kernel hyperparameter, i.e. , \lambda in $k(x,y)=\exp(-\lambda \Vert{x-y}\Vert²)$ in the kernel function. This repository contains an effective algorithmn that calculates an approximate entire solution path of the objective function with respect to the hyperparameter within the interval $[2^{-10},2^{10}]$ without numerical issues by which the exact algorithms suffer. More details can be found in the corresponding paper.

For the matrix calculation the library Eigen was used in combination with OpenMP. The backend solver is libSVM

The program reads the problem description in the default libSVM format. See the documented source for more information or run the compiled program without any parameters to get help.

weblinks