[9112] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
| 3 | * Copyright (C) 2002-2012 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
| 4 | *
|
---|
| 5 | * This file is part of HeuristicLab.
|
---|
| 6 | *
|
---|
| 7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
| 8 | * it under the terms of the GNU General Public License as published by
|
---|
| 9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
| 10 | * (at your option) any later version.
|
---|
| 11 | *
|
---|
| 12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 15 | * GNU General Public License for more details.
|
---|
| 16 | *
|
---|
| 17 | * You should have received a copy of the GNU General Public License
|
---|
| 18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
| 19 | */
|
---|
| 20 | #endregion
|
---|
| 21 |
|
---|
| 22 | using System;
|
---|
| 23 | using System.Collections.Generic;
|
---|
| 24 | using System.Linq;
|
---|
| 25 | using HeuristicLab.Common;
|
---|
| 26 | using HeuristicLab.Core;
|
---|
| 27 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 28 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 29 | using ILNumerics;
|
---|
| 30 |
|
---|
| 31 | namespace HeuristicLab.Algorithms.DataAnalysis {
|
---|
| 32 | /// <summary>
|
---|
| 33 | /// Represents a Gaussian process model.
|
---|
| 34 | /// </summary>
|
---|
| 35 | [StorableClass]
|
---|
| 36 | [Item("TunedGaussianProcessModel", "Gaussian process model implemented using ILNumerics.")]
|
---|
| 37 | public sealed class TunedGaussianProcessModel : NamedItem, IGaussianProcessModel {
|
---|
| 38 | [Storable]
|
---|
| 39 | private double negativeLogLikelihood;
|
---|
| 40 | public double NegativeLogLikelihood {
|
---|
| 41 | get { return negativeLogLikelihood; }
|
---|
| 42 | }
|
---|
| 43 |
|
---|
| 44 | [Storable]
|
---|
| 45 | private double[] hyperparameterGradients;
|
---|
| 46 | public double[] HyperparameterGradients {
|
---|
| 47 | get {
|
---|
| 48 | var copy = new double[hyperparameterGradients.Length];
|
---|
| 49 | Array.Copy(hyperparameterGradients, copy, copy.Length);
|
---|
| 50 | return copy;
|
---|
| 51 | }
|
---|
| 52 | }
|
---|
| 53 |
|
---|
| 54 | [Storable]
|
---|
| 55 | private ICovarianceFunction covarianceFunction;
|
---|
| 56 | public ICovarianceFunction CovarianceFunction {
|
---|
| 57 | get { return covarianceFunction; }
|
---|
| 58 | }
|
---|
| 59 | [Storable]
|
---|
| 60 | private IMeanFunction meanFunction;
|
---|
| 61 | public IMeanFunction MeanFunction {
|
---|
| 62 | get { return meanFunction; }
|
---|
| 63 | }
|
---|
| 64 | [Storable]
|
---|
| 65 | private string targetVariable;
|
---|
| 66 | public string TargetVariable {
|
---|
| 67 | get { return targetVariable; }
|
---|
| 68 | }
|
---|
| 69 | [Storable]
|
---|
| 70 | private string[] allowedInputVariables;
|
---|
| 71 | public string[] AllowedInputVariables {
|
---|
| 72 | get { return allowedInputVariables; }
|
---|
| 73 | }
|
---|
| 74 |
|
---|
| 75 | [Storable]
|
---|
| 76 | private double[] alpha;
|
---|
| 77 |
|
---|
| 78 |
|
---|
| 79 | [Storable]
|
---|
| 80 | private double sqrSigmaNoise;
|
---|
| 81 | public double SigmaNoise {
|
---|
| 82 | get { return Math.Sqrt(sqrSigmaNoise); }
|
---|
| 83 | }
|
---|
| 84 |
|
---|
| 85 | [Storable]
|
---|
| 86 | private double[] meanParameter;
|
---|
| 87 | [Storable]
|
---|
| 88 | private double[] covarianceParameter;
|
---|
| 89 |
|
---|
| 90 | [Storable]
|
---|
| 91 | private double[] l;
|
---|
| 92 |
|
---|
| 93 | [Storable]
|
---|
| 94 | private double[,] x;
|
---|
| 95 | [Storable]
|
---|
| 96 | private Scaling inputScaling;
|
---|
| 97 |
|
---|
| 98 |
|
---|
| 99 | [StorableConstructor]
|
---|
| 100 | private TunedGaussianProcessModel(bool deserializing) : base(deserializing) { }
|
---|
| 101 | private TunedGaussianProcessModel(TunedGaussianProcessModel original, Cloner cloner)
|
---|
| 102 | : base(original, cloner) {
|
---|
| 103 | this.meanFunction = cloner.Clone(original.meanFunction);
|
---|
| 104 | this.covarianceFunction = cloner.Clone(original.covarianceFunction);
|
---|
| 105 | this.inputScaling = cloner.Clone(original.inputScaling);
|
---|
| 106 | this.negativeLogLikelihood = original.negativeLogLikelihood;
|
---|
| 107 | this.targetVariable = original.targetVariable;
|
---|
| 108 | this.sqrSigmaNoise = original.sqrSigmaNoise;
|
---|
| 109 | if (original.meanParameter != null) {
|
---|
| 110 | this.meanParameter = (double[])original.meanParameter.Clone();
|
---|
| 111 | }
|
---|
| 112 | if (original.covarianceParameter != null) {
|
---|
| 113 | this.covarianceParameter = (double[])original.covarianceParameter.Clone();
|
---|
| 114 | }
|
---|
| 115 |
|
---|
| 116 | // shallow copies of arrays because they cannot be modified
|
---|
| 117 | this.allowedInputVariables = original.allowedInputVariables;
|
---|
| 118 | this.alpha = original.alpha;
|
---|
| 119 | this.l = original.l;
|
---|
| 120 | this.x = original.x;
|
---|
| 121 | }
|
---|
| 122 | public TunedGaussianProcessModel(Dataset ds, string targetVariable, IEnumerable<string> allowedInputVariables, IEnumerable<int> rows,
|
---|
| 123 | IEnumerable<double> hyp, IMeanFunction meanFunction, ICovarianceFunction covarianceFunction)
|
---|
| 124 | : base() {
|
---|
| 125 | this.name = ItemName;
|
---|
| 126 | this.description = ItemDescription;
|
---|
| 127 | this.meanFunction = (IMeanFunction)meanFunction.Clone();
|
---|
| 128 | this.covarianceFunction = (ICovarianceFunction)covarianceFunction.Clone();
|
---|
| 129 | this.targetVariable = targetVariable;
|
---|
| 130 | this.allowedInputVariables = allowedInputVariables.ToArray();
|
---|
| 131 |
|
---|
| 132 |
|
---|
| 133 | int nVariables = this.allowedInputVariables.Length;
|
---|
| 134 | meanParameter = hyp
|
---|
| 135 | .Take(this.meanFunction.GetNumberOfParameters(nVariables))
|
---|
| 136 | .ToArray();
|
---|
| 137 |
|
---|
| 138 | covarianceParameter = hyp.Skip(this.meanFunction.GetNumberOfParameters(nVariables))
|
---|
| 139 | .Take(this.covarianceFunction.GetNumberOfParameters(nVariables))
|
---|
| 140 | .ToArray();
|
---|
| 141 | sqrSigmaNoise = Math.Exp(2.0 * hyp.Last());
|
---|
| 142 |
|
---|
| 143 | CalculateModel(ds, rows);
|
---|
| 144 |
|
---|
| 145 | //var cmpModel = new GaussianProcessModel(ds, targetVariable, allowedInputVariables, rows, hyp,
|
---|
| 146 | // (IMeanFunction)meanFunction.Clone(),
|
---|
| 147 | // (ICovarianceFunction)covarianceFunction.Clone());
|
---|
| 148 | //if (!cmpModel.NegativeLogLikelihood.IsAlmost(NegativeLogLikelihood) ||
|
---|
| 149 | // !cmpModel.HyperparameterGradients.Sum().IsAlmost(HyperparameterGradients.Sum())
|
---|
| 150 | // )
|
---|
| 151 | // throw new ArgumentException();
|
---|
| 152 | }
|
---|
| 153 |
|
---|
| 154 | private void CalculateModel(Dataset ds, IEnumerable<int> rows) {
|
---|
| 155 | inputScaling = new Scaling(ds, allowedInputVariables, rows);
|
---|
| 156 | x = AlglibUtil.PrepareAndScaleInputMatrix(ds, allowedInputVariables, rows, inputScaling);
|
---|
| 157 | var y = ds.GetDoubleValues(targetVariable, rows);
|
---|
| 158 | ILNumerics.Settings.MaxNumberThreads = Environment.ProcessorCount;
|
---|
| 159 | using (ILScope.Enter()) {
|
---|
| 160 | int n = x.GetLength(0);
|
---|
| 161 |
|
---|
| 162 | // calculate means and covariances
|
---|
| 163 | var mean = meanFunction.GetParameterizedMeanFunction(meanParameter, Enumerable.Range(0, x.GetLength(1)));
|
---|
| 164 | double[] m = Enumerable.Range(0, x.GetLength(0))
|
---|
| 165 | .Select(r => mean.Mean(x, r))
|
---|
| 166 | .ToArray();
|
---|
| 167 |
|
---|
| 168 | var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, null);
|
---|
| 169 | ILArray<double> myL = ILMath.zeros<double>(n, n);
|
---|
| 170 |
|
---|
| 171 | for (int i = 0; i < n; i++) {
|
---|
| 172 | for (int j = i; j < n; j++) {
|
---|
| 173 | double c = cov.Covariance(x, i, j) / sqrSigmaNoise;
|
---|
| 174 | if (i == j) {
|
---|
| 175 | myL.SetValue(c + 1, i, j);
|
---|
| 176 | } else {
|
---|
| 177 | myL.SetValue(c, j, i);
|
---|
| 178 | myL.SetValue(c, i, j);
|
---|
| 179 | }
|
---|
| 180 | }
|
---|
| 181 | }
|
---|
| 182 |
|
---|
| 183 | // cholesky decomposition
|
---|
[9124] | 184 | ILArray<double> chol;
|
---|
| 185 | try {
|
---|
| 186 | chol = ILNumerics.ILMath.chol(myL, false);
|
---|
| 187 | }
|
---|
| 188 | catch (Exception e) {
|
---|
| 189 | throw new ArgumentException("covariance matrix not positive definite", e);
|
---|
| 190 | }
|
---|
[9112] | 191 | if (chol == null || chol.IsEmpty || !chol.Size.Equals(myL.Size))
|
---|
| 192 | throw new ArgumentException("covariance matrix not positive definite");
|
---|
| 193 | this.l = new double[n * n];
|
---|
| 194 | chol.ExportValues(ref l);
|
---|
| 195 | // calculate sum of diagonal elements for likelihood
|
---|
| 196 |
|
---|
| 197 | double diagSum = ILMath.log(ILMath.diag(chol)).Sum();
|
---|
| 198 |
|
---|
| 199 | // solve for alpha
|
---|
| 200 | ILArray<double> ym = ILMath.array(y.Zip(m, (a, b) => a - b).ToArray());
|
---|
| 201 |
|
---|
| 202 | MatrixProperties lowerTriProps = MatrixProperties.LowerTriangular;
|
---|
| 203 | MatrixProperties upperTriProps = MatrixProperties.UpperTriangular;
|
---|
| 204 | ILArray<double> alpha;
|
---|
| 205 | try {
|
---|
| 206 | ILArray<double> t1 = ILMath.linsolve(chol.T, ym, ref lowerTriProps);
|
---|
| 207 | alpha = ILMath.linsolve(chol, t1, ref upperTriProps) / sqrSigmaNoise;
|
---|
| 208 | }
|
---|
| 209 | catch (Exception e) {
|
---|
| 210 | throw new ArgumentException("covariance matrix is not positive definite", e);
|
---|
| 211 | }
|
---|
| 212 | if (alpha == null || alpha.IsEmpty) throw new ArgumentException();
|
---|
| 213 | this.alpha = new double[alpha.Length];
|
---|
| 214 | alpha.ExportValues(ref this.alpha);
|
---|
| 215 |
|
---|
| 216 | negativeLogLikelihood = 0.5 * (double)ILMath.multiply(ym.T, alpha) + diagSum + (n / 2.0) * Math.Log(2.0 * Math.PI * sqrSigmaNoise);
|
---|
| 217 |
|
---|
| 218 | // derivatives
|
---|
| 219 | int nAllowedVariables = x.GetLength(1);
|
---|
| 220 |
|
---|
| 221 | ILArray<double> lCopy;
|
---|
| 222 | try {
|
---|
| 223 | ILArray<double> t2 = ILMath.linsolve(chol.T, ILMath.eye<double>(n, n), ref lowerTriProps);
|
---|
| 224 | lCopy = ILMath.linsolve(chol, t2, ref upperTriProps) / sqrSigmaNoise - ILMath.multiply(alpha, alpha.T);
|
---|
| 225 | }
|
---|
| 226 | catch (Exception e) {
|
---|
| 227 | throw new ArgumentException("covariance matrix is not positive definite", e);
|
---|
| 228 | }
|
---|
| 229 |
|
---|
| 230 | double noiseGradient = sqrSigmaNoise * ILMath.sumall(ILMath.diag(lCopy)).GetValue(0, 0);
|
---|
| 231 |
|
---|
| 232 | double[] meanGradients = new double[meanFunction.GetNumberOfParameters(nAllowedVariables)];
|
---|
| 233 | for (int k = 0; k < meanGradients.Length; k++) {
|
---|
| 234 | ILArray<double> meanGrad =
|
---|
| 235 | ILMath.array(
|
---|
| 236 | Enumerable.Range(0, alpha.Length)
|
---|
| 237 | .Select(r => mean.Gradient(x, r, k))
|
---|
| 238 | .ToArray());
|
---|
| 239 | meanGradients[k] = -(double)ILMath.multiply(meanGrad.T, alpha);
|
---|
| 240 | }
|
---|
| 241 |
|
---|
| 242 | double[] covGradients = new double[covarianceFunction.GetNumberOfParameters(nAllowedVariables)];
|
---|
| 243 | if (covGradients.Length > 0) {
|
---|
| 244 | for (int i = 0; i < n; i++) {
|
---|
| 245 | for (int j = 0; j < i; j++) {
|
---|
| 246 | var g = cov.CovarianceGradient(x, i, j).ToArray();
|
---|
| 247 | for (int k = 0; k < covGradients.Length; k++) {
|
---|
| 248 | covGradients[k] += lCopy.GetValue(i, j) * g[k];
|
---|
| 249 | }
|
---|
| 250 | }
|
---|
| 251 |
|
---|
| 252 | var gDiag = cov.CovarianceGradient(x, i, i).ToArray();
|
---|
| 253 | for (int k = 0; k < covGradients.Length; k++) {
|
---|
| 254 | // diag
|
---|
| 255 | covGradients[k] += 0.5 * lCopy.GetValue(i, i) * gDiag[k];
|
---|
| 256 | }
|
---|
| 257 | }
|
---|
| 258 | }
|
---|
| 259 |
|
---|
| 260 | hyperparameterGradients =
|
---|
| 261 | meanGradients
|
---|
| 262 | .Concat(covGradients)
|
---|
| 263 | .Concat(new double[] { noiseGradient }).ToArray();
|
---|
| 264 | }
|
---|
| 265 | }
|
---|
| 266 |
|
---|
| 267 |
|
---|
| 268 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 269 | return new TunedGaussianProcessModel(this, cloner);
|
---|
| 270 | }
|
---|
| 271 |
|
---|
| 272 | // is called by the solution creator to set all parameter values of the covariance and mean function
|
---|
| 273 | // to the optimized values (necessary to make the values visible in the GUI)
|
---|
| 274 | public void FixParameters() {
|
---|
| 275 | covarianceFunction.SetParameter(covarianceParameter);
|
---|
| 276 | meanFunction.SetParameter(meanParameter);
|
---|
| 277 | covarianceParameter = new double[0];
|
---|
| 278 | meanParameter = new double[0];
|
---|
| 279 | }
|
---|
| 280 |
|
---|
| 281 | #region IRegressionModel Members
|
---|
| 282 | public IEnumerable<double> GetEstimatedValues(Dataset dataset, IEnumerable<int> rows) {
|
---|
| 283 | return GetEstimatedValuesHelper(dataset, rows);
|
---|
| 284 | }
|
---|
| 285 | public GaussianProcessRegressionSolution CreateRegressionSolution(IRegressionProblemData problemData) {
|
---|
| 286 | return new GaussianProcessRegressionSolution(this, new RegressionProblemData(problemData));
|
---|
| 287 | }
|
---|
| 288 | IRegressionSolution IRegressionModel.CreateRegressionSolution(IRegressionProblemData problemData) {
|
---|
| 289 | return CreateRegressionSolution(problemData);
|
---|
| 290 | }
|
---|
| 291 | #endregion
|
---|
| 292 |
|
---|
| 293 |
|
---|
| 294 | private IEnumerable<double> GetEstimatedValuesHelper(Dataset dataset, IEnumerable<int> rows) {
|
---|
| 295 | var newX = AlglibUtil.PrepareAndScaleInputMatrix(dataset, allowedInputVariables, rows, inputScaling);
|
---|
| 296 | int newN = newX.GetLength(0);
|
---|
| 297 | int n = x.GetLength(0);
|
---|
| 298 | var Ks = new double[newN, n];
|
---|
| 299 | var mean = meanFunction.GetParameterizedMeanFunction(meanParameter, Enumerable.Range(0, newX.GetLength(1)));
|
---|
| 300 | var ms = Enumerable.Range(0, newX.GetLength(0))
|
---|
| 301 | .Select(r => mean.Mean(newX, r))
|
---|
| 302 | .ToArray();
|
---|
| 303 | var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, null);
|
---|
| 304 | for (int i = 0; i < newN; i++) {
|
---|
| 305 | for (int j = 0; j < n; j++) {
|
---|
| 306 | Ks[i, j] = cov.CrossCovariance(x, newX, j, i);
|
---|
| 307 | }
|
---|
| 308 | }
|
---|
| 309 |
|
---|
| 310 | return Enumerable.Range(0, newN)
|
---|
| 311 | .Select(i => ms[i] + Util.ScalarProd(Util.GetRow(Ks, i), alpha));
|
---|
| 312 | }
|
---|
| 313 |
|
---|
| 314 | public IEnumerable<double> GetEstimatedVariance(Dataset dataset, IEnumerable<int> rows) {
|
---|
| 315 | var newX = AlglibUtil.PrepareAndScaleInputMatrix(dataset, allowedInputVariables, rows, inputScaling);
|
---|
| 316 | int newN = newX.GetLength(0);
|
---|
| 317 | int n = x.GetLength(0);
|
---|
| 318 | if (newN == 0) return Enumerable.Empty<double>();
|
---|
| 319 |
|
---|
| 320 | var kss = new double[newN];
|
---|
| 321 | ILArray<double> sWKs = ILMath.zeros(n, newN);
|
---|
| 322 | var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, null);
|
---|
| 323 |
|
---|
| 324 | // for stddev
|
---|
| 325 | for (int i = 0; i < newN; i++)
|
---|
| 326 | kss[i] = cov.Covariance(newX, i, i);
|
---|
| 327 |
|
---|
| 328 | for (int i = 0; i < newN; i++) {
|
---|
| 329 | for (int j = 0; j < n; j++) {
|
---|
| 330 | sWKs.SetValue(cov.CrossCovariance(x, newX, j, i) / Math.Sqrt(sqrSigmaNoise), j, i);
|
---|
| 331 | }
|
---|
| 332 | }
|
---|
| 333 |
|
---|
| 334 | // for stddev
|
---|
| 335 | var lowerTriangular = MatrixProperties.LowerTriangular;
|
---|
| 336 | ILArray<double> V = ILMath.linsolve(ILMath.array(l, n, n).T, sWKs, ref lowerTriangular);
|
---|
| 337 |
|
---|
| 338 | // alglib.ablas.rmatrixlefttrsm(n, newN, l, 0, 0, false, false, 0, ref sWKs, 0, 0);
|
---|
| 339 |
|
---|
| 340 | for (int i = 0; i < newN; i++) {
|
---|
| 341 | double sumV = (double)ILMath.multiply(V[ILMath.full, i].T, V[ILMath.full, i]);
|
---|
| 342 | kss[i] -= sumV;
|
---|
| 343 | if (kss[i] < 0) kss[i] = 0;
|
---|
| 344 | }
|
---|
| 345 | return kss;
|
---|
| 346 | }
|
---|
| 347 | }
|
---|
| 348 | }
|
---|