1 /**
2 * Copyright (c) 2011, The University of Southampton and the individual contributors.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without modification,
6 * are permitted provided that the following conditions are met:
7 *
8 * * Redistributions of source code must retain the above copyright notice,
9 * this list of conditions and the following disclaimer.
10 *
11 * * Redistributions in binary form must reproduce the above copyright notice,
12 * this list of conditions and the following disclaimer in the documentation
13 * and/or other materials provided with the distribution.
14 *
15 * * Neither the name of the University of Southampton nor the names of its
16 * contributors may be used to endorse or promote products derived from this
17 * software without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
21 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
22 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
23 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
24 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
25 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
26 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
28 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29 */
30 package org.openimaj.math.matrix.algorithm.ica;
31
32 import org.openimaj.math.matrix.MatrixUtils;
33 import org.openimaj.util.pair.IndependentPair;
34
35 import Jama.Matrix;
36
37 public abstract class IndependentComponentAnalysis {
38 public abstract Matrix getSignalToInterferenceMatrix();
39
40 public abstract Matrix getDemixingMatrix();
41
42 public abstract Matrix getIndependentComponentMatrix();
43
44 /**
45 * Estimate the independent components of the given data array. Each row
46 * corresponds to an observation with the number of dimensions equal to the
47 * number of columns.
48 *
49 * @param data
50 * the data
51 */
52 public void estimateComponents(double[][] data) {
53 estimateComponents(new Matrix(data));
54 }
55
56 /**
57 * Estimate the independent components of the given data array. Each row
58 * corresponds to an observation with the number of dimensions equal to the
59 * number of columns.
60 *
61 * @param data
62 * the data
63 */
64 public void estimateComponents(Matrix data) {
65 final IndependentPair<Matrix, double[]> p = computeMeanCentre(data);
66 final Matrix meanCentredX = p.firstObject();
67 final double[] mean = p.getSecondObject();
68
69 final IndependentPair<Matrix, Matrix> p2 = decorrelate(meanCentredX);
70 final Matrix Z = p2.firstObject();
71 final Matrix CC = p2.secondObject();
72 estimateComponentsWhitened(Z, mean, meanCentredX, CC);
73 }
74
75 private IndependentPair<Matrix, Matrix> decorrelate(Matrix meanCentredX) {
76 final Matrix C = MatrixUtils.covariance(meanCentredX.transpose());
77 final Matrix CC = MatrixUtils.invSqrtSym(C);
78 return IndependentPair.pair(CC.times(meanCentredX), CC);
79 }
80
81 private IndependentPair<Matrix, double[]> computeMeanCentre(Matrix m) {
82 final double[][] data = m.getArray();
83
84 final double[] mean = new double[data.length];
85
86 for (int j = 0; j < data.length; j++)
87 for (int i = 0; i < data[0].length; i++)
88 mean[j] += data[j][i];
89
90 for (int i = 0; i < data.length; i++)
91 mean[i] /= data[0].length;
92
93 final Matrix mat = new Matrix(data.length, data[0].length);
94 final double[][] matdat = mat.getArray();
95
96 for (int j = 0; j < data.length; j++)
97 for (int i = 0; i < data[0].length; i++)
98 matdat[j][i] = (data[j][i] - mean[j]);
99
100 return IndependentPair.pair(mat, mean);
101 }
102
103 /**
104 * Estimate the IC's from the given decorrelated (mean-centred and whitened)
105 * data matrix, Z.
106 *
107 * @param Z
108 * the whitened data; one observation per row
109 * @param mean
110 * the mean of each dimension
111 * @param X
112 * the mean-centered data; one observation per row
113 */
114 protected abstract void estimateComponentsWhitened(Matrix Z, double[] mean, Matrix X, Matrix CC);
115 }