Package edu.pitt.csb.mgm

Class Mgm

java.lang.Object

edu.pitt.csb.mgm.ConvexProximal

edu.pitt.csb.mgm.Mgm

All Implemented Interfaces:

IGraphSearch

public class Mgm
extends ConvexProximal
implements IGraphSearch

Implementation of Lee and Hastie's (2012) pseudolikelihood method for learning Mixed Gaussian-Categorical Graphical Models Created by ajsedgewick on 7/15/15.

Version:

$Id: $Id

Author:

josephramsey

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	Mgm.MGMParams	The parameters of the MGM model.

Constructor Summary

Constructors

Constructor	Description
Mgm(cern.colt.matrix.DoubleMatrix2D x, cern.colt.matrix.DoubleMatrix2D y, List<Node> variables, int[] l, double[] lambda)	Constructor for Mgm.
Mgm(DataSet ds, double[] lambda)	Constructor for Mgm.

Method Summary

Modifier and Type	Method	Description
cern.colt.matrix.DoubleMatrix2D	adjMatFromMGM()	Converts MGM to matrix of doubles.
static cern.colt.matrix.DoubleMatrix1D	flatten(cern.colt.matrix.DoubleMatrix2D m)	flatten.
long	getElapsedTime()	Return time of execution for learning.
Graph	graphFromMGM()	Converts MGM object to Graph object with edges if edge parameters are non-zero.
void	learn(double epsilon, int iterLimit)	Learn MGM traditional way with objective function tolerance.
void	learnEdges(int iterLimit)	Learn MGM using edge convergence using default 3 iterations of no edge changes.
void	learnEdges(int iterLimit, int edgeChangeTol)	Learn MGM using edge convergence using edgeChangeTol (see ProximalGradient for documentation).
static void	main(String[] args)	main.
double	nonSmooth(double t, cern.colt.matrix.DoubleMatrix1D X, cern.colt.matrix.DoubleMatrix1D pX)	Calculates the non-smooth value for a given parameter and input vectors.
double	nonSmoothValue(cern.colt.matrix.DoubleMatrix1D parIn)	Calculates the non-smooth value for the given input vector.
cern.colt.matrix.DoubleMatrix1D	proximalOperator(double t, cern.colt.matrix.DoubleMatrix1D X)	Applies proximal operator on the given input vector with a positive parameter.
Graph	search()	Simple search command for GraphSearch implementation.
void	setParams(Mgm.MGMParams newParams)	Setter for the field params.
double	smooth(cern.colt.matrix.DoubleMatrix1D parIn, cern.colt.matrix.DoubleMatrix1D gradOutVec)	Smooth method calculates the smooth loss and gradient given input parameters.
cern.colt.matrix.DoubleMatrix1D	smoothGradient(cern.colt.matrix.DoubleMatrix1D parIn)	Calculates the smooth gradient for a given input vector.
double	smoothValue(cern.colt.matrix.DoubleMatrix1D parIn)	Calculate the smooth value of the given input vector.
static cern.colt.matrix.DoubleMatrix2D	upperTri(cern.colt.matrix.DoubleMatrix2D mat, int di)	upperTri.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

Mgm

public Mgm(cern.colt.matrix.DoubleMatrix2D x,
 cern.colt.matrix.DoubleMatrix2D y,
 List<Node> variables,
 int[] l,
 double[] lambda)

Constructor for Mgm.

Parameters:

x - a DoubleMatrix2D object

y - a DoubleMatrix2D object

variables - a List object

l - an array of objects

lambda - an array of objects

Mgm

public Mgm(DataSet ds,
 double[] lambda)

Constructor for Mgm.

Parameters:

ds - a DataSet object

lambda - an array of objects

Method Details

flatten

public static cern.colt.matrix.DoubleMatrix1D flatten(cern.colt.matrix.DoubleMatrix2D m)

flatten.

Parameters:

m - a DoubleMatrix2D object

Returns:

a DoubleMatrix1D object

upperTri

public static cern.colt.matrix.DoubleMatrix2D upperTri(cern.colt.matrix.DoubleMatrix2D mat,
 int di)

upperTri.

Parameters:

mat - a DoubleMatrix2D object

di - a int

Returns:

a DoubleMatrix2D object

main

public static void main(String[] args)

main.

Parameters:

args - an array of String objects

setParams

public void setParams(Mgm.MGMParams newParams)

Setter for the field params.

Parameters:

newParams - a Mgm.MGMParams object

smoothValue

public double smoothValue(cern.colt.matrix.DoubleMatrix1D parIn)

Calculate the smooth value of the given input vector.

Parameters:

parIn - The input vector.

Returns:

The smooth value.

smooth

public double smooth(cern.colt.matrix.DoubleMatrix1D parIn,
 cern.colt.matrix.DoubleMatrix1D gradOutVec)

Smooth method calculates the smooth loss and gradient given input parameters.

Overrides:

smooth in class ConvexProximal

Parameters:

parIn - input Vector

gradOutVec - gradient of g(X)

Returns:

the smooth loss

nonSmoothValue

public double nonSmoothValue(cern.colt.matrix.DoubleMatrix1D parIn)

Calculates the non-smooth value for the given input vector.

Parameters:

parIn - the input vector

Returns:

the non-smooth value

smoothGradient

public cern.colt.matrix.DoubleMatrix1D smoothGradient(cern.colt.matrix.DoubleMatrix1D parIn)

Calculates the smooth gradient for a given input vector.

Parameters:

parIn - the input vector

Returns:

the smooth gradient

proximalOperator

public cern.colt.matrix.DoubleMatrix1D proximalOperator(double t,
 cern.colt.matrix.DoubleMatrix1D X)

Applies proximal operator on the given input vector with a positive parameter.

Parameters:

t - the positive parameter for proximal operator

X - the input vector

Returns:

the result of applying proximal operator on the input vector

Throws:

IllegalArgumentException - if t is not positive

nonSmooth

public double nonSmooth(double t,
 cern.colt.matrix.DoubleMatrix1D X,
 cern.colt.matrix.DoubleMatrix1D pX)

Calculates the non-smooth value for a given parameter and input vectors.

Overrides:

nonSmooth in class ConvexProximal

Parameters:

t - the positive parameter for the prox operator

X - the input vector

pX - the vector solution to prox_t(X)

Returns:

the non-smooth value

learn

public void learn(double epsilon,
 int iterLimit)

Learn MGM traditional way with objective function tolerance. Recommended for inference applications that need accurate pseudolikelihood

Parameters:

epsilon - tolerance in change of objective function

iterLimit - iteration limit

learnEdges

public void learnEdges(int iterLimit)

Learn MGM using edge convergence using default 3 iterations of no edge changes. Recommended when we only care about edge existence.

Parameters:

iterLimit - a int

learnEdges

public void learnEdges(int iterLimit,
 int edgeChangeTol)

Learn MGM using edge convergence using edgeChangeTol (see ProximalGradient for documentation). Recommended when we only care about edge existence.

Parameters:

iterLimit - a int

edgeChangeTol - a int

graphFromMGM

public Graph graphFromMGM()

Converts MGM object to Graph object with edges if edge parameters are non-zero. Loses all edge param information

Returns:

a Graph object

adjMatFromMGM

public cern.colt.matrix.DoubleMatrix2D adjMatFromMGM()

Converts MGM to matrix of doubles. uses 2-norm to combine c-d edge parameters into single value and f-norm for d-d edge parameters.

Returns:

a DoubleMatrix2D object

search

public Graph search()

Simple search command for GraphSearch implementation. Uses default edge convergence, 1000 iter limit.

Specified by:

search in interface IGraphSearch

Returns:

a Graph object

getElapsedTime

public long getElapsedTime()

Return time of execution for learning.

Returns:

a long