|
QuasiNewtonMethod * | quasi_newton_method_pointer = nullptr |
|
Tensor< type, 1 > | old_parameters |
|
Tensor< type, 1 > | parameters_difference |
|
Tensor< type, 1 > | parameters_increment |
|
Tensor< type, 1 > | old_gradient |
|
Tensor< type, 1 > | gradient_difference |
|
Tensor< type, 2 > | inverse_hessian |
|
Tensor< type, 2 > | old_inverse_hessian |
|
Tensor< type, 1 > | old_inverse_hessian_dot_gradient_difference |
|
Index | epoch = 0 |
|
Tensor< type, 0 > | training_slope |
|
type | learning_rate = type(0) |
|
type | old_learning_rate = type(0) |
|
Tensor< type, 1 > | potential_parameters |
|
Tensor< type, 1 > | training_direction |
|
type | initial_learning_rate = type(0) |
|
Definition at line 184 of file quasi_newton_method.h.
◆ QuasiNewtonMehtodData() [1/2]
◆ QuasiNewtonMehtodData() [2/2]
◆ ~QuasiNewtonMehtodData()
◆ print()
◆ set()
◆ epoch
◆ gradient_difference
Tensor<type, 1> gradient_difference |
◆ inverse_hessian
Tensor<type, 2> inverse_hessian |
◆ learning_rate
type learning_rate = type(0) |
◆ old_gradient
Tensor<type, 1> old_gradient |
◆ old_inverse_hessian
Tensor<type, 2> old_inverse_hessian |
◆ old_inverse_hessian_dot_gradient_difference
Tensor<type, 1> old_inverse_hessian_dot_gradient_difference |
◆ old_learning_rate
type old_learning_rate = type(0) |
◆ old_parameters
Tensor<type, 1> old_parameters |
◆ parameters_difference
Tensor<type, 1> parameters_difference |
◆ parameters_increment
Tensor<type, 1> parameters_increment |
◆ quasi_newton_method_pointer
◆ training_slope
Tensor<type, 0> training_slope |
The documentation for this struct was generated from the following file: