NormalRBFuncC - Computes normal parameters of a set of unlabbelled data using EM alg

User Documentation

Centre for Vision, Speech & Signal Processing

Comments:
This function estimates a PDF as a Gausian mixture. It can be constructed from a set of data points. The parameters of the gaussians are then estimated using the Expectation Maximisation algorithm (Dempster et al (1977) J. Royal Statistical Society Series B, 39:1-38).

Parent Classes:

NumFuncC

Methods:

NormalRBFuncC()

empty constructor

NormalRBFuncC(UIntT mixes)

Construct RBF from unlabelled data using a specified number of mixes

mixes	The number of gaussians you require in your mixture

NormalRBFuncC(istream & in)

Construct from the input stream

NormalRBFuncC(const SArray1dC<VecMatC> & params,const VectorC & weights)

Construct from user given class parameters

params	array of distribution statistics. Each VecMatC contains the mean and covariance for the class corresponding to its index value in the array.
weights	vector containing weighting coefficient for each Gaussian mixture.

NormalRBFuncC(const Array1dC<VecMatC> & params,const VectorC & weights)

Construct from user given class parameters

params	array of distribution statistics. Each VecMatC contains the mean and covariance for the class corresponding to its index value in the array.
weights	vector containing weighting coefficient for each Gaussian mixture.

NormalRBFuncC(const VectorC & mean,const MatrixC & cov)

Construct a single component mixture model with mean and covariance

mean	mean vector.
cov	covariance matrix.

NormalRBFuncC Copy() const

Makes a deep copy and is virtual

Methods exclusive to RBF

void Join(const NormalRBFuncC & rbf)

Joins two models together. It will also normalise the weights

void Normalise()

Normalises the weights

void Save(ostream & out) const

Save the model to stream

void Save(FilenameC & out) const

Save the model to file

RealT NegLogLikelihood(const DListC<VectorC> & data) const

Calculate the likelihood of the data given the model

data	the data

BooleanT ExpectationMaximization(const DListC<VectorC> & data)

Performs the standard batch-mode EM algorithm on the data in the list

data	training vectors

BooleanT EM(const DListC<VectorC> & data)

Performs the standard batch-mode EM algorithm on the data in the list (slower version of above)

data	training vectors

BooleanT ProbPcaEM(const DListC<VectorC> & data,UIntT dimen)

Performs the Probabilistic PCA version of EM Algorithm

data	training vectors

BooleanT Update(const DListC<VectorC> & data,RealT alpha = 0.1,BooleanT freqSen = FALSE)

Performs the on-line version of the EM algorithm on the data in the list. The model is updated after each point. Each point in the list is only presented once. This routine should be used when a temporal stream of data is available. The batch EM is computationally safer!

data	training vectors
alpha	learning rate
freqSen	use frequency sensitive version of alg. (only use if you know what it is and what it does).

UIntT Size() const

Number of Gaussian mixtures used to represent distribution

UIntT Dim() const

Number of Gaussian mixtures used to represent distribution

VectorC Weights() const

Access to the mixing coefficients

void Weights(const VectorC & w)

Set the mixing coefficients

SArray1dC<VecMatC> Parameters() const

Access to the means and covariances of the mixture

void Parameters(const SArray1dC<VecMatC> & p)

Set the covariance matrices and means of the mixture

SArray1dC<MatrixC> getInvCov() const

Access to the inverse of the covariances matrices

void setInvCov(const SArray1dC<MatrixC> & p)

Set the inverse of the covariance matrix

SArray1dC<RealT> getDet() const

Access to the determinants

void setDet(const SArray1dC<RealT> & p)

Set the determinants

Array1dC<VecMatC> Array() const

Array access to parameters

RealT Posterior(const VectorC & X,UIntT index)

Calculate normalised posterior prob for X and mixture index

VectorC PosteriorProb(const VectorC & X)

Calculate normalised posterior probs for all inputs

BooleanT Init() const

Check status of initialisation flag

void Init(BooleanT flag)

Set status of initiliasation flag

BooleanT FullCovMatrix() const

Toggle between full and diagonal covariance matrices

void FullCovMatrix(BooleanT flag)

Set covariances matrix to FULL or DIAG

RealT L2norm(const NormalRBFuncC & rbf) const

Set covariances matrix to FULL or DIAG

DListC<NormalRBFuncC> Group() const

Group the components using PEAKS method

DListC<NormalRBFuncC> Group(const SArray1dC<VecMatC> & compStats) const

Group the components using STATS method

DListC<NormalRBFuncC> GroupOR(const SArray1dC<VecMatC> & compStats) const

Group the components using both PEAKS then STATS

DListC<NormalRBFuncC> GroupEuclid(const SArray1dC<VectorC> & groupsMean) const

Group the components according to Euclidean distance of the components to the desired mean

DListC<NormalRBFuncC> GroupMahal(const SArray1dC<VecMatC> & groupsParams) const

Group the components according to Mahalanobis distance of the components to the global model components

DListC<NormalRBFuncC> GroupBattachar(const SArray1dC<VecMatC> & groupsParams) const

Group the components according to Battacharya distance of the components to the global model components

NumLabelC GroupLabel(const DListC<NormalRBFuncC> & group,const VectorC & X) const

Labeling of the vector X using groups of NormalRBFunc

DListC<NumLabelC> GroupLabel(const DListC<NormalRBFuncC> & group,const DListC<VectorC> & L) const

Labeling of the vector X using groups of NormalRBFunc

RealT PredictedProb(const WindowT & win) const

Calculate the predicted probability in a given window Uses vegas if FULL covariances matrices used

RealT PredictedProb(const WindowT & win,IntT ncall,IntT itmx) const

Calculate the predicted probability in a given window Uses vegas if FULL covariances matrices used

void InsertComponent(const VectorC & mn,const MatrixC & cov,const RealT wt)

Calculate the predicted probability in a given window Uses vegas if FULL covariances matrices used

ostream & operator<<(ostream & s,const NormalRBFuncC & out)

output stream operator

istream & operator>>(istream & s,NormalRBFuncC & in)

input stream operator

#include "amma/Num/NumFunc.hh"

BooleanT IsA(const DPEntityC & pb): Check if an object is a NumFuncC.
NumFuncC Copy() const: Makes a deep copy
NumFuncC & operator=(const NumFuncC & oth): Assignment operator
VectorC Evaluate(const VectorC & X) const: Evaluates function Y=f(X)
DListC<VectorC> Evaluate(const DListC<VectorC> & listX) const: Evaluates function Y=f(X) for a list of X
VectorC operator()(const VectorC & X) const: Evaluates function Y=f(X)
MatrixC Jacobian(const VectorC & X) const: Evaluates Jacobian df(X)/dX
Note that this is not const=0 since the base class will calculate a numerical estimate of the Jacobian using differences if the member function is not overloaded in derived classes to provide an analytical solution.
const StringC GetInfo() const: Gets string describing object
const StringC GetName() const: Gets type name of the object
UIntT SizeX() const: Input vector dimension
UIntT SizeY() const: Output vector dimension
BooleanT Save(ostream & out) const: Writes object to stream, cna be loaded using constructor

#include "amma/DP/Process.hh"

VectorC Apply(const VectorC & dat): Apply operation.
NB. This may become constant in the future, but the situation isn't clear at the moment.
IntT ApplyArray(const SArray1dC<VectorC> & in,SArray1dC<VectorC> & out): Streaming version.
const DPProcessC<VectorC,VectorC> operator=(const DPProcessC<VectorC,VectorC> & in): Assignment operator
const DPProcessC<VectorC,VectorC> Copy() const: Make a copy of this process. Some type defs.
const type_info & InputType() const: Get input type.
const type_info & OutputType() const: Get input type.
BooleanT IsStateless() const: Is operation stateless ?
DPProcessBaseBodyC::ProcTypeT OpType() const: Operation type lossy/lossless.
const DPProcessBaseC & operator=(const DPProcessBaseC & oth): Assignment operator.

#include "amma/DP/Entity.hh"

const DPEntityC & operator=(const DPEntityC & dat): Assignment.
DPEntityBodyC & Body(): Access body.
const DPEntityBodyC & Body() const: Access body.
BooleanT Save(ostream & out) const

#include "amma/RCHandleA.hh"

RCAbstractC Abstract(): Create an abstract handle to this object.
BooleanT IsHandleType(const DT &) const: Is handle of given type ?
void CheckHandleType(const DT & dummy) const: Check handle type. Throw an expception if not.
BooleanT Save(ostream & out) const: Save to ostream.
RCHandleAC<DPEntityBodyC> Copy() const: Creat a deep copy of this object.
protected:

#include "amma/RCHandle.hh"

BooleanT IsValid() const: Check its a valid handle.
BooleanT IsValidObject() const: Check its a valid handle, and that the object is pointed to is also valid.
RCHandleC<DPEntityBodyC> & operator=(const RCHandleC<DPEntityBodyC> & oth): Assignment.
BooleanT operator==(const RCHandleC<DPEntityBodyC> & oth) const: Comparison operator.
BooleanT operator!=(const RCHandleC<DPEntityBodyC> & oth) const: Comparison operator.
UIntT Hash() const: Hash function.x
BooleanT IsConst() const: Is object constant ?
BooleanT IsNotConst() const: Is object not constant ?
void SetConst(void) const: Lock the object.
This is const as a convience though it actual modified the object, often object you wish to ensure are constant already have const set. protected:
void Invalidate(): Turn this into an invalid handle.
DPEntityBodyC & Body(): Direct access to body.
const DPEntityBodyC & Body() const: Constant access to body.

Programmer:Kieron Messer, Documentation by CxxDoc: Tue Mar 20 10:48:08 2001

NormalRBFuncC

Computes normal parameters of a set of unlabbelled data using EM alg

include	"amma/Num/NormalRBFunc.hh"
User Level:	Default
Library:	NumRBF
Example:	doNormalRbf.cc
Section:	Pattern Recognition.Radial Basis Functions
In Scope:	std