mic::neural_nets::optimization::AdaGrad< eT > Class Template Reference

Update using AdaGrad - adaptive gradient descent. More...

#include <AdaGrad.hpp>

Inheritance diagram for mic::neural_nets::optimization::AdaGrad< eT >:

Collaboration diagram for mic::neural_nets::optimization::AdaGrad< eT >:

Public Member Functions
	AdaGrad (size_t rows_, size_t cols_, eT eps_=1e-8)
mic::types::MatrixPtr< eT >	calculateUpdate (mic::types::MatrixPtr< eT > x_, mic::types::MatrixPtr< eT > dx_, eT learning_rate_)
Public Member Functions inherited from mic::neural_nets::optimization::OptimizationFunction< eT >
	OptimizationFunction ()
virtual	~OptimizationFunction ()
	Virtual destructor - empty. More...
virtual void	update (mic::types::MatrixPtr< eT > p_, mic::types::MatrixPtr< eT > dp_, eT learning_rate_, eT decay_=0.0)
virtual void	update (mic::types::MatrixPtr< eT > p_, mic::types::MatrixPtr< eT > x_, mic::types::MatrixPtr< eT > y_, eT learning_rate_=0.001)

Protected Attributes
eT	eps
	Smoothing term that avoids division by zero. More...
mic::types::MatrixPtr< eT >	G
	Sum of all of the squares of the gradients up to time t ("diagonal matrix"). More...
mic::types::MatrixPtr< eT >	delta
	Calculated update. More...

Detailed Description

template<typename eT = float>
class mic::neural_nets::optimization::AdaGrad< eT >

Update using AdaGrad - adaptive gradient descent.

Author

tkornuta

Definition at line 39 of file AdaGrad.hpp.

Constructor & Destructor Documentation

template<typename eT = float>

mic::neural_nets::optimization::AdaGrad< eT >::AdaGrad ( size_t  rows_, size_t  cols_, eT  eps_ = 1e-8  )

inline

Constructor. Sets dimensions and eps (default=1e-8).

Parameters

rows_	Number of rows of the updated matrix/its gradient.
cols_	Number of columns of the updated matrix/its gradient.

Definition at line 47 of file AdaGrad.hpp.

References mic::neural_nets::optimization::AdaGrad< eT >::delta, and mic::neural_nets::optimization::AdaGrad< eT >::G.

Member Function Documentation

template<typename eT = float>

mic::types::MatrixPtr<eT> mic::neural_nets::optimization::AdaGrad< eT >::calculateUpdate ( mic::types::MatrixPtr< eT >  x_, mic::types::MatrixPtr< eT >  dx_, eT  learning_rate_  )

inlinevirtual

Performs update according to the AdaGrad update rule.

Parameters

x_	Pointer to the current matrix.
dx_	Pointer to current gradient of that matrix.
learning_rate_	Learning rate (default=0.001).

Implements mic::neural_nets::optimization::OptimizationFunction< eT >.

Definition at line 63 of file AdaGrad.hpp.

References mic::neural_nets::optimization::AdaGrad< eT >::delta, mic::neural_nets::optimization::AdaGrad< eT >::eps, and mic::neural_nets::optimization::AdaGrad< eT >::G.

Member Data Documentation

template<typename eT = float>

mic::types::MatrixPtr<eT> mic::neural_nets::optimization::AdaGrad< eT >::delta

protected

Calculated update.

Definition at line 87 of file AdaGrad.hpp.

Referenced by mic::neural_nets::optimization::AdaGrad< eT >::AdaGrad(), and mic::neural_nets::optimization::AdaGrad< eT >::calculateUpdate().

template<typename eT = float>

eT mic::neural_nets::optimization::AdaGrad< eT >::eps

protected

Smoothing term that avoids division by zero.

Definition at line 81 of file AdaGrad.hpp.

Referenced by mic::neural_nets::optimization::AdaGrad< eT >::calculateUpdate().

template<typename eT = float>

mic::types::MatrixPtr<eT> mic::neural_nets::optimization::AdaGrad< eT >::G

protected

Sum of all of the squares of the gradients up to time t ("diagonal matrix").

Definition at line 84 of file AdaGrad.hpp.

Referenced by mic::neural_nets::optimization::AdaGrad< eT >::AdaGrad(), and mic::neural_nets::optimization::AdaGrad< eT >::calculateUpdate().

---

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

• src/optimization/AdaGrad.hpp