Thrill
0.1
|
Definition at line 99 of file stochastic_gradient_descent.hpp.
#include <stochastic_gradient_descent.hpp>
Public Member Functions | |
StochasticGradientDescent (size_t num_iterations, double mini_batch_fraction, double step_size, double tolerance) | |
Vector | optimize (const DIA< DataPoint< Vector > > &input_points, const Vector &initial_weights) |
do the actual computation More... | |
Private Member Functions | |
bool | is_converged (Vector &old, Vector &curr, double tolerance) |
Private Attributes | |
double | mini_batch_fraction |
size_t | num_iterations |
double | step_size |
double | tolerance |
|
inline |
Definition at line 102 of file stochastic_gradient_descent.hpp.
Definition at line 167 of file stochastic_gradient_descent.hpp.
References Vector< D, Type >::Distance(), max(), and Vector< D, Type >::Norm().
|
inline |
do the actual computation
Definition at line 111 of file stochastic_gradient_descent.hpp.
References LeastSquaresGradient< Vector >::Compute(), SumResult< Vector >::count, DataPoint< Vector >::data, SumResult< Vector >::grad, DataPoint< Vector >::label, LOG1, and Vector< D, double >::Make().
|
private |
Definition at line 163 of file stochastic_gradient_descent.hpp.
|
private |
Definition at line 162 of file stochastic_gradient_descent.hpp.
|
private |
Definition at line 164 of file stochastic_gradient_descent.hpp.
|
private |
Definition at line 165 of file stochastic_gradient_descent.hpp.