SHOGUN  4.1.0
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Modules Pages
List of all members | Public Member Functions
DescendUpdater Class Referenceabstract

Detailed Description

This is a base class for descend update.

The class give the interface used in descend-based minimizer.

Given a target variable, \(w\), and its negative descend direction \(g\), the class will update \(w\) based on \(g\) (eg, subtracting \(g\))

Note that an example of \(g\) is to simply use the gradient wrt \(w\).

Definition at line 49 of file DescendUpdater.h.

Inheritance diagram for DescendUpdater:
Inheritance graph
[legend]

Public Member Functions

virtual void update_variable (SGVector< float64_t > variable_reference, SGVector< float64_t > negative_descend_direction, float64_t learning_rate)=0
 
virtual void update_context (CMinimizerContext *context)=0
 
virtual void load_from_context (CMinimizerContext *context)=0
 

Member Function Documentation

virtual void load_from_context ( CMinimizerContext context)
pure virtual

Load the given context object to restore mutable variables

This method will be called by FirstOrderMinimizer::load_from_context(CMinimizerContext* context)

Parameters
contexta context object

Implemented in DescendUpdaterWithCorrection, AdaDeltaUpdater, RmsPropUpdater, and AdaGradUpdater.

virtual void update_context ( CMinimizerContext context)
pure virtual

Update a context object to store mutable variables used in descend update

This method will be called by FirstOrderMinimizer::save_to_context()

Parameters
contexta context object

Implemented in DescendUpdaterWithCorrection, AdaDeltaUpdater, RmsPropUpdater, and AdaGradUpdater.

virtual void update_variable ( SGVector< float64_t variable_reference,
SGVector< float64_t negative_descend_direction,
float64_t  learning_rate 
)
pure virtual

Update the target variable based on the given negative descend direction

Note that this method will update the target variable in place. This method will be called by FirstOrderMinimizer::minimize()

Parameters
variable_referencea reference of the target variable
negative_descend_directionthe negative descend direction given the current value
learning_ratelearning rate

Implemented in AdaDeltaUpdater, RmsPropUpdater, AdaGradUpdater, and DescendUpdaterWithCorrection.


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

SHOGUN Machine Learning Toolbox - Documentation