MetaParametersLWPR Class Reference

Meta-parameters for the Locally Weighted Projection Regression (LWPR) function approximator. More...

#include <MetaParametersLWPR.hpp>

Inheritance diagram for MetaParametersLWPR:

Collaboration diagram for MetaParametersLWPR:

Public Member Functions
	MetaParametersLWPR (int expected_input_dim, Eigen::VectorXd init_D=Eigen::VectorXd::Ones(1), double w_gen=0.2, double w_prune=0.8, bool update_D=true, double init_alpha=1.0, double penalty=1.0, bool diag_only=true, bool use_meta=false, double meta_rate=1.0, std::string kernel_name=std::string("Gaussian"))
	Constructor for the algorithmic meta-parameters of the LWPR function approximator. More...
MetaParametersLWPR *	clone (void) const
	Return a pointer to a deep copy of the MetaParameters object. More...
std::string	toString (void) const
	Returns a string representation of the object. More...
Public Member Functions inherited from MetaParameters
	MetaParameters (int expected_input_dim)
	Constructor. More...
virtual	~MetaParameters (void)
	Virtual destructor, because this is a base class.
int	getExpectedInputDim (void) const
	The expected dimensionality of the input data. More...
virtual int	getExpectedOutputDim (void) const
	The expected dimensionality of the output data. More...

Friends
class	FunctionApproximatorLWPR
class	boost::serialization::access
	Give boost serialization access to private members. More...

Additional Inherited Members
Protected Member Functions inherited from MetaParameters
	MetaParameters (void)
	Default constructor. More...

Detailed Description

Meta-parameters for the Locally Weighted Projection Regression (LWPR) function approximator.

Definition at line 37 of file MetaParametersLWPR.hpp.

Constructor & Destructor Documentation

MetaParametersLWPR	(	int	expected_input_dim,
		Eigen::VectorXd	init_D = Eigen::VectorXd::Ones(1),
		double	w_gen = 0.2,
		double	w_prune = 0.8,
		bool	update_D = true,
		double	init_alpha = 1.0,
		double	penalty = 1.0,
		bool	diag_only = true,
		bool	use_meta = false,
		double	meta_rate = 1.0,
		std::string	kernel_name = std::string("Gaussian")
	)

Constructor for the algorithmic meta-parameters of the LWPR function approximator.

The meaning of these parameters is explained here: http://wcms.inf.ed.ac.uk/ipab/slmc/research/software-lwpr

Short howto:

• Set update_D=false, diag_only=true, use_meta=false
• Tune init_D, and then w_gen and w_prune
• Set update_D=true, then tune init_alpha, then penalty
• Set diag_only=false, see if it helps, if so re-tune init_alpha if necessary
• Set use_meta=true, tune meta_rate (I never do this...)

Parameters

[in]	expected_input_dim	Expected dimensionality of the input data
[in]	init_D	Removing/adding kernels: Width of a kernel when it is newly placed. Smaller * values mean wider kernels.
[in]	w_gen	Removing/adding kernels: Threshold for adding a kernel.
[in]	w_prune	Removing/adding kernels: Threshold for pruning a kernel.
[in]	update_D	Updating existing kernels: whether to update kernels
[in]	init_alpha	Updating existing kernels: rate at which kernels are updated
[in]	penalty	Updating existing kernels: regularization term. Higher penalty means less kernels.
[in]	diag_only	Whether to update only the diagonal of the covariance matrix of the kernel, or the full matrix.
[in]	use_meta	Meta-learning of kernel update rate: whether meta-learning is enabled
[in]	meta_rate	Meta-learning of kernel update rate: meta-learning rate
[in]	kernel_name	Removing/adding kernels: Type of kernels

Definition at line 48 of file MetaParametersLWPR.cpp.

:
  MetaParameters(expected_input_dim),
  init_D_(init_D), w_gen_(w_gen), w_prune_(w_prune),
  update_D_(update_D), init_alpha_(init_alpha), penalty_(penalty), diag_only_(diag_only),
  use_meta_(use_meta), meta_rate_(meta_rate), kernel_name_(kernel_name)
{
  assert(init_D_.size()==expected_input_dim);
  assert(w_gen_>0.0 && w_gen_<1.0);
  assert(w_prune_>0.0 && w_prune_<1.0);
  assert(w_gen_<w_prune_);
}

Member Function Documentation

MetaParametersLWPR * clone ( void  ) const

virtual

Return a pointer to a deep copy of the MetaParameters object.

Returns

Pointer to a deep copy

Implements MetaParameters.

Definition at line 66 of file MetaParametersLWPR.cpp.

{
  return new MetaParametersLWPR(
    getExpectedInputDim(), 
    init_D_, w_gen_, w_prune_,
    update_D_, init_alpha_, penalty_, diag_only_,
    use_meta_, meta_rate_, kernel_name_
  );  
}

Here is the call graph for this function:

string toString ( void  ) const

virtual

Returns a string representation of the object.

Returns

A string representation of the object.

Implements MetaParameters.

Definition at line 76 of file MetaParametersLWPR.cpp.

                                              {
  RETURN_STRING_FROM_BOOST_SERIALIZATION_XML("MetaParametersLWPR");
}

Friends And Related Function Documentation

friend class boost::serialization::access

friend

Give boost serialization access to private members.

Definition at line 113 of file MetaParametersLWPR.hpp.

---

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

• MetaParametersLWPR.hpp
• MetaParametersLWPR.cpp