org.openimaj.ml.linear.data

Class BiconvexDataGenerator

java.lang.Object

org.openimaj.ml.linear.data.BiconvexDataGenerator

All Implemented Interfaces:

DataGenerator<gov.sandia.cognition.math.matrix.Matrix,gov.sandia.cognition.math.matrix.Matrix>, MatrixDataGenerator<gov.sandia.cognition.math.matrix.Matrix>

public class BiconvexDataGenerator
extends Object
implements MatrixDataGenerator<gov.sandia.cognition.math.matrix.Matrix>

Data generated from a biconvex system of the form: Y_n,t = U_:,t^T . X_n . W_t + rand() Note that each n'th instance of Y can have values for T tasks. The parameter matricies U and W can be independant of tasks or note. The amount of random noise added can be controlled The sparcity of U, W and X can be controlled, the sparcity of U and W is consistent per generator. The sparcity of X changes per generation.

Author:

Sina Samangooei (ss@ecs.soton.ac.uk)

Constructor Summary

Constructors

Constructor and Description
BiconvexDataGenerator() Generates a biconvex data generator.
BiconvexDataGenerator(int nusers, int nfeatures, int ntasks, double sparcity, double xsparcity, boolean indw, boolean indu, int seed, double noise)

Method Summary

Modifier and Type	Method and Description
org.openimaj.util.pair.Pair<gov.sandia.cognition.math.matrix.Matrix>	generate()
gov.sandia.cognition.math.matrix.Matrix	getU()
gov.sandia.cognition.math.matrix.Matrix	getW()

Methods inherited from class java.lang.Object

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

Constructor Detail

BiconvexDataGenerator

public BiconvexDataGenerator()

Generates a biconvex data generator.

BiconvexDataGenerator

public BiconvexDataGenerator(int nusers,
                             int nfeatures,
                             int ntasks,
                             double sparcity,
                             double xsparcity,
                             boolean indw,
                             boolean indu,
                             int seed,
                             double noise)

Parameters:

nusers - The number of users (U is users x tasks, X is features x users)

nfeatures - The number of features (W is features x tasks)

ntasks - The number of tasks (Y is 1 x tasks)

sparcity - The chance that a row of U or W is zeros

xsparcity - The chance that a column of U or W is zeros

indw - If true, there is a column of W per task

indu - If true, there is a column of U per task

seed - If greater than or equal to zero, the rng backing this generator is seeded

noise - The random noise added to Y has random values for each Y ranging from -noise to noise

Method Detail

generate

public org.openimaj.util.pair.Pair<gov.sandia.cognition.math.matrix.Matrix> generate()

Specified by:

generate in interface DataGenerator<gov.sandia.cognition.math.matrix.Matrix,gov.sandia.cognition.math.matrix.Matrix>

Specified by:

generate in interface MatrixDataGenerator<gov.sandia.cognition.math.matrix.Matrix>

Returns:

The X and Y of this generator's Y = F(X)

getU

public gov.sandia.cognition.math.matrix.Matrix getU()

getW

public gov.sandia.cognition.math.matrix.Matrix getW()