AdamUpdate Class Reference

Adam is an optimizer that computes individual adaptive learning rates for different parameters from estimates of first and second moments of the gradients as given in the section 7 of the following paper. More...

Public Member Functions
	AdamUpdate (const double epsilon=1e-8, const double beta1=0.9, const double beta2=0.999)
	Construct the Adam update policy with the given parameters. More...
double	Beta1 () const
	Get the smoothing parameter. More...
double &	Beta1 ()
	Modify the smoothing parameter. More...
double	Beta2 () const
	Get the second moment coefficient. More...
double &	Beta2 ()
	Modify the second moment coefficient. More...
double	Epsilon () const
	Get the value used to initialise the squared gradient parameter. More...
double &	Epsilon ()
	Modify the value used to initialise the squared gradient parameter. More...
void	Initialize (const size_t rows, const size_t cols)
	The Initialize method is called by SGD Optimizer method before the start of the iteration update process. More...
void	Update (arma::mat &iterate, const double stepSize, const arma::mat &gradient)
	Update step for Adam. More...

Detailed Description

Adam is an optimizer that computes individual adaptive learning rates for different parameters from estimates of first and second moments of the gradients as given in the section 7 of the following paper.

For more information, see the following.

@article{Kingma2014,
  author  = {Diederik P. Kingma and Jimmy Ba},
  title   = {Adam: {A} Method for Stochastic Optimization},
  journal = {CoRR},
  year    = {2014},
  url     = {http://arxiv.org/abs/1412.6980}
}

Definition at line 42 of file adam_update.hpp.

Constructor & Destructor Documentation

AdamUpdate()

AdamUpdate ( const double  epsilon = 1e-8, const double  beta1 = 0.9, const double  beta2 = 0.999  )

inline

Construct the Adam update policy with the given parameters.

Parameters

epsilon	The epsilon value used to initialise the squared gradient parameter.
beta1	The smoothing parameter.
beta2	The second moment coefficient.

Definition at line 53 of file adam_update.hpp.

Member Function Documentation

Beta1() [1/2]

double Beta1 ( ) const

inline

Get the smoothing parameter.

Definition at line 116 of file adam_update.hpp.

Beta1() [2/2]

double& Beta1 ( )

inline

Modify the smoothing parameter.

Definition at line 118 of file adam_update.hpp.

Beta2() [1/2]

double Beta2 ( ) const

inline

Get the second moment coefficient.

Definition at line 121 of file adam_update.hpp.

Beta2() [2/2]

double& Beta2 ( )

inline

Modify the second moment coefficient.

Definition at line 123 of file adam_update.hpp.

Epsilon() [1/2]

double Epsilon ( ) const

inline

Get the value used to initialise the squared gradient parameter.

Definition at line 111 of file adam_update.hpp.

Epsilon() [2/2]

double& Epsilon ( )

inline

Modify the value used to initialise the squared gradient parameter.

Definition at line 113 of file adam_update.hpp.

Initialize()

void Initialize ( const size_t  rows, const size_t  cols  )

inline

The Initialize method is called by SGD Optimizer method before the start of the iteration update process.

Parameters

rows	Number of rows in the gradient matrix.
cols	Number of columns in the gradient matrix.

Definition at line 71 of file adam_update.hpp.

Update()

void Update ( arma::mat &  iterate, const double  stepSize, const arma::mat &  gradient  )

inline

Update step for Adam.

Parameters

iterate	Parameters that minimize the function.
stepSize	Step size to be used for the given iteration.
gradient	The gradient matrix.

It should be noted that the term, m / (arma::sqrt(v) + eps), in the following expression is an approximation of the following actual term; m / (arma::sqrt(v) + (arma::sqrt(biasCorrection2) * eps).

Definition at line 84 of file adam_update.hpp.

---

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

• src/mlpack/core/optimizers/adam/adam_update.hpp