ML::SVM::LinearSVMParams Struct Reference

#include <linear.hpp>

Collaboration diagram for ML::SVM::LinearSVMParams:

Public Types
enum	Penalty { L1 , L2 }
enum	Loss { HINGE , SQUARED_HINGE , EPSILON_INSENSITIVE , SQUARED_EPSILON_INSENSITIVE }

Public Attributes
Penalty	penalty = L2
Loss	loss = HINGE
bool	fit_intercept = true
bool	penalized_intercept = false
bool	probability = false
int	max_iter = 1000
int	linesearch_max_iter = 100
int	lbfgs_memory = 5
int	verbose = 0
double	C = 1.0
double	grad_tol = 0.0001
double	change_tol = 0.00001
double	epsilon = 0.0

Member Enumeration Documentation

Loss

enum ML::SVM::LinearSVMParams::Loss

The loss function.

Enumerator
HINGE	max(1 - y_i x_i w, 0)
SQUARED_HINGE	max(1 - y_i x_i w, 0)^2
EPSILON_INSENSITIVE	max(|y_i - x_i w| - epsilon, 0)
SQUARED_EPSILON_INSENSITIVE	max(|y_i - x_i w| - epsilon, 0)^2

Penalty

enum ML::SVM::LinearSVMParams::Penalty

The regularization term.

Enumerator
L1	Abs. value of the weights: sum |w|
L2	Squared value of the weights: sum w^2

Member Data Documentation

C

double ML::SVM::LinearSVMParams::C = 1.0

The constant scaling factor of the main term in the loss function. (You can also think of that as the inverse factor of the penalty term).

change_tol

double ML::SVM::LinearSVMParams::change_tol = 0.00001

The threshold on the function change for the underlying QN solver.

epsilon

double ML::SVM::LinearSVMParams::epsilon = 0.0

The epsilon-sensitivity parameter (applicable to the SVM-regression (SVR) loss functions).

fit_intercept

bool ML::SVM::LinearSVMParams::fit_intercept = true

Whether to fit the bias term.

grad_tol

double ML::SVM::LinearSVMParams::grad_tol = 0.0001

The threshold on the gradient for the underlying QN solver.

lbfgs_memory

int ML::SVM::LinearSVMParams::lbfgs_memory = 5

Number of vectors approximating the hessian for the underlying QN solver (l-bfgs).

linesearch_max_iter

int ML::SVM::LinearSVMParams::linesearch_max_iter = 100

Maximum number of linesearch (inner loop) iterations for the underlying QN solver.

loss

Loss ML::SVM::LinearSVMParams::loss = HINGE

The loss function.

max_iter

int ML::SVM::LinearSVMParams::max_iter = 1000

Maximum number of iterations for the underlying QN solver.

penalized_intercept

bool ML::SVM::LinearSVMParams::penalized_intercept = false

When true, the bias term is treated the same way as other data features. Enabling this feature forces an extra copying the input data X.

penalty

Penalty ML::SVM::LinearSVMParams::penalty = L2

The regularization term.

probability

bool ML::SVM::LinearSVMParams::probability = false

Whether to estimate probabilities using Platt scaling (applicable to SVC).

verbose

int ML::SVM::LinearSVMParams::verbose = 0

Triggers extra output when greater than zero.

---

The documentation for this struct was generated from the following file:

• include/cuml/svm/linear.hpp