MiniDNN::SGD Class Reference

#include <SGD.h>

Inheritance diagram for MiniDNN::SGD:

Public Member Functions
void	update (ConstAlignedMapVec &dvec, AlignedMapVec &vec)
Public Member Functions inherited from MiniDNN::Optimizer
virtual void	reset ()

Public Attributes
Scalar	m_lrate
Scalar	m_decay

Detailed Description

The Stochastic Gradient Descent (SGD) algorithm

Definition at line 16 of file SGD.h.

Member Function Documentation

update()

void MiniDNN::SGD::update ( ConstAlignedMapVec &  dvec, AlignedMapVec &  vec  )

inlinevirtual

Update the parameter vector using its gradient

It is assumed that the memory addresses of dvec and vec do not change during the training process. This is used to implement optimization algorithms that have "memories". See the AdaGrad algorithm for an example.

Parameters

dvec	The gradient of the parameter. Read-only
vec	On entering, the current parameter vector. On exit, the updated parameters.

Implements MiniDNN::Optimizer.

Definition at line 31 of file SGD.h.

---

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

• Optimizer/SGD.h