InverseScalingLearningRate Class Reference

Detailed Description

The implements the inverse scaling learning rate.

The learning rate is computed in the following way:

\[ \frac{\eta_0}{{(a+b \times iter)}^k} \]

where \(\eta_0\) is the initial learning rate, \(a\) is the intercept term, \(b\) is the slope term, \(iter\) is the number of times to call get_learning_rate(), and \(k\) is the exponent term.

Definition at line 53 of file InverseScalingLearningRate.h.

Inheritance diagram for InverseScalingLearningRate:

Public Member Functions
	InverseScalingLearningRate ()
virtual	~InverseScalingLearningRate ()
virtual float64_t	get_learning_rate (int32_t iter_counter)
virtual void	set_initial_learning_rate (float64_t initial_learning_rate)
virtual void	set_exponent (float64_t exponent)
virtual void	set_slope (float64_t slope)
virtual void	set_intercept (float64_t intercept)
virtual void	update_context (CMinimizerContext *context)
virtual void	load_from_context (CMinimizerContext *context)

Protected Attributes
float64_t	m_exponent
float64_t	m_slope
float64_t	m_intercept
float64_t	m_initial_learning_rate

Constructor & Destructor Documentation

InverseScalingLearningRate

(

)

Definition at line 57 of file InverseScalingLearningRate.h.

virtual ~InverseScalingLearningRate ( )

virtual

Definition at line 60 of file InverseScalingLearningRate.h.

Member Function Documentation

virtual float64_t get_learning_rate ( int32_t  iter_counter )

virtual

Get the learning rate for descent direction

Parameters

iter_counter

the number of iterations

Returns

the learning rate (A.K.A step size/length)

Implements LearningRate.

Definition at line 67 of file InverseScalingLearningRate.h.

virtual void load_from_context ( CMinimizerContext *  context )

virtual

Return a context object which stores mutable variables Usually it is used in serialization.

Returns

a context object

Implements LearningRate.

Definition at line 129 of file InverseScalingLearningRate.h.

virtual void set_exponent ( float64_t  exponent )

virtual

Set the exponent term

Parameters

exponent

exponent term should be positive

Definition at line 88 of file InverseScalingLearningRate.h.

virtual void set_initial_learning_rate ( float64_t  initial_learning_rate )

virtual

Set the initial learning rate

Parameters

initial_learning_rate

initial_learning_rate must be positive

Definition at line 77 of file InverseScalingLearningRate.h.

virtual void set_intercept ( float64_t  intercept )

virtual

Set the intercept term

Parameters

intercept

intercept term should be positive

Definition at line 108 of file InverseScalingLearningRate.h.

virtual void set_slope ( float64_t  slope )

virtual

Set the slope term

Parameters

slope

slope term should be positive

Definition at line 98 of file InverseScalingLearningRate.h.

virtual void update_context ( CMinimizerContext *  context )

virtual

Update a context object to store mutable variables

Parameters

context

a context object

Implements LearningRate.

Definition at line 119 of file InverseScalingLearningRate.h.

Member Data Documentation

float64_t m_exponent

protected

exponent

Definition at line 135 of file InverseScalingLearningRate.h.

float64_t m_initial_learning_rate

protected

init_learning_rate

Definition at line 141 of file InverseScalingLearningRate.h.

float64_t m_intercept

protected

intercept

Definition at line 139 of file InverseScalingLearningRate.h.

float64_t m_slope

protected

slope

Definition at line 137 of file InverseScalingLearningRate.h.

---

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

• InverseScalingLearningRate.h