#include <svm_parameter.h>
Collaboration diagram for ML::SVM::SvmParameter:
Public Attributes | |
| double | C |
| Penalty term C. More... | |
| double | cache_size |
| int | max_iter |
| int | nochange_steps |
| double | tol |
| Tolerance used to stop fitting. More... | |
| rapids_logger::level_enum | verbosity |
| Print information about training. More... | |
| double | epsilon |
| epsilon parameter for epsilon-SVR More... | |
| SvmType | svmType |
Detailed Description
Numerical input parameters for an SVM.
There are several parameters that control how long we train. The training stops if:
- max_iter iterations are reached. If you pass -1, then max_diff = 100 * n_rows
- the diff becomes less the tol
- the diff is changing less then 0.001*tol in nochange_steps consecutive outer iterations.
Member Data Documentation
◆ C
| double ML::SVM::SvmParameter::C |
Penalty term C.
◆ cache_size
| double ML::SVM::SvmParameter::cache_size |
kernel cache size in MiB
◆ epsilon
| double ML::SVM::SvmParameter::epsilon |
epsilon parameter for epsilon-SVR
◆ max_iter
| int ML::SVM::SvmParameter::max_iter |
maximum number of outer SMO iterations. Use -1 to let the SMO solver set a default value (100*n_rows).
◆ nochange_steps
| int ML::SVM::SvmParameter::nochange_steps |
◆ svmType
| SvmType ML::SVM::SvmParameter::svmType |
◆ tol
| double ML::SVM::SvmParameter::tol |
Tolerance used to stop fitting.
◆ verbosity
| rapids_logger::level_enum ML::SVM::SvmParameter::verbosity |
Print information about training.
The documentation for this struct was generated from the following file:
- include/cuml/svm/svm_parameter.h
