Changeset 8982 for trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/GaussianProcessModel.cs

Timestamp:

12/01/12 19:02:47 (12 years ago)

Author:

gkronber

Message:

#1902: removed class HyperParameter and changed implementations of covariance and mean functions to remove the parameter value caching and event handlers for parameter caching. Instead it is now possible to create the actual covariance and mean functions as Func from templates and specified parameter values. The instances of mean and covariance functions configured in the GUI are actually templates where the structure and fixed parameters can be specified.

File:

• trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/GaussianProcessModel.cs (modified) (10 diffs)

trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/GaussianProcessModel.cs

@@ -81 +81 @@
 
     [Storable]
+    private double[] meanParameter;
+    [Storable]
+    private double[] covarianceParameter;
+
+    [Storable]
     private double[,] l;
 
@@ -99 +104 @@
       this.targetVariable = original.targetVariable;
       this.sqrSigmaNoise = original.sqrSigmaNoise;
+      if (original.meanParameter != null) {
+        this.meanParameter = (double[])original.meanParameter.Clone();
+      }
+      if (original.covarianceParameter != null) {
+        this.covarianceParameter = (double[])original.covarianceParameter.Clone();
+      }
 
       // shallow copies of arrays because they cannot be modified
@@ -118 +129 @@
 
       int nVariables = this.allowedInputVariables.Length;
-      this.meanFunction.SetParameter(hyp
+      meanParameter = hyp
         .Take(this.meanFunction.GetNumberOfParameters(nVariables))
-        .ToArray());
-      this.covarianceFunction.SetParameter(hyp.Skip(this.meanFunction.GetNumberOfParameters(nVariables))
-        .Take(this.covarianceFunction.GetNumberOfParameters(nVariables))
-        .ToArray());
+        .ToArray();
+
+      covarianceParameter = hyp.Skip(this.meanFunction.GetNumberOfParameters(nVariables))
+                                             .Take(this.covarianceFunction.GetNumberOfParameters(nVariables))
+                                             .ToArray();
       sqrSigmaNoise = Math.Exp(2.0 * hyp.Last());
 
@@ -138 +150 @@
 
       // calculate means and covariances
-      double[] m = meanFunction.GetMean(x);
+      var mean = meanFunction.GetParameterizedMeanFunction(meanParameter, Enumerable.Range(0, x.GetLength(1)));
+      double[] m = Enumerable.Range(0, x.GetLength(0))
+        .Select(r => mean.Mean(x, r))
+        .ToArray();
+
+      var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, Enumerable.Range(0, x.GetLength(1)));
       for (int i = 0; i < n; i++) {
         for (int j = i; j < n; j++) {
-          l[j, i] = covarianceFunction.GetCovariance(x, i, j) / sqrSigmaNoise;
+          l[j, i] = cov.Covariance(x, i, j) / sqrSigmaNoise;
           if (j == i) l[j, i] += 1.0;
         }
       }
+
 
       // cholesky decomposition
@@ -181 +199 @@
 
       double[] meanGradients = new double[meanFunction.GetNumberOfParameters(nAllowedVariables)];
-      for (int i = 0; i < meanGradients.Length; i++) {
-        var meanGrad = meanFunction.GetGradients(i, x);
-        meanGradients[i] = -Util.ScalarProd(meanGrad, alpha);
+      for (int k = 0; k < meanGradients.Length; k++) {
+        var meanGrad = Enumerable.Range(0, alpha.Length)
+        .Select(r => mean.Gradient(x, r, k));
+        meanGradients[k] = -Util.ScalarProd(meanGrad, alpha);
       }
 
@@ -190 +209 @@
         for (int i = 0; i < n; i++) {
           for (int j = 0; j < i; j++) {
-            var g = covarianceFunction.GetGradient(x, i, j).ToArray();
+            var g = cov.CovarianceGradient(x, i, j).ToArray();
             for (int k = 0; k < covGradients.Length; k++) {
               covGradients[k] += lCopy[i, j] * g[k];
@@ -196 +215 @@
           }
 
-          var gDiag = covarianceFunction.GetGradient(x, i, i).ToArray();
+          var gDiag = cov.CovarianceGradient(x, i, i).ToArray();
           for (int k = 0; k < covGradients.Length; k++) {
             // diag
@@ -216 +235 @@
     }
 
+    // is called by the solution creator to set all parameter values of the covariance and mean function
+    // to the optimized values (necessary to make the values visible in the GUI)
+    public void FixParameters() {
+      covarianceFunction.SetParameter(covarianceParameter);
+      meanFunction.SetParameter(meanParameter);
+      covarianceParameter = new double[0];
+      meanParameter = new double[0];
+    }
+
     #region IRegressionModel Members
     public IEnumerable<double> GetEstimatedValues(Dataset dataset, IEnumerable<int> rows) {
@@ -234 +262 @@
       int n = x.GetLength(0);
       var Ks = new double[newN, n];
-      var ms = meanFunction.GetMean(newX);
+      var mean = meanFunction.GetParameterizedMeanFunction(meanParameter, Enumerable.Range(0, newX.GetLength(1)));
+      var ms = Enumerable.Range(0, newX.GetLength(0))
+      .Select(r => mean.Mean(newX, r))
+      .ToArray();
+      var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, Enumerable.Range(0, newX.GetLength(1)));
       for (int i = 0; i < newN; i++) {
         for (int j = 0; j < n; j++) {
-          Ks[i, j] = covarianceFunction.GetCrossCovariance(x, newX, j, i);
+          Ks[i, j] = cov.CrossCovariance(x, newX, j, i);
         }
       }
@@ -252 +284 @@
       var kss = new double[newN];
       double[,] sWKs = new double[n, newN];
+      var cov = covarianceFunction.GetParameterizedCovarianceFunction(covarianceParameter, Enumerable.Range(0, newX.GetLength(1)));
 
       // for stddev 
       for (int i = 0; i < newN; i++)
-        kss[i] = covarianceFunction.GetCovariance(newX, i, i);
+        kss[i] = cov.Covariance(newX, i, i);
 
       for (int i = 0; i < newN; i++) {
         for (int j = 0; j < n; j++) {
-          sWKs[j, i] = covarianceFunction.GetCrossCovariance(x, newX, j, i) / Math.Sqrt(sqrSigmaNoise);
+          sWKs[j, i] = cov.CrossCovariance(x, newX, j, i) / Math.Sqrt(sqrSigmaNoise);
         }
       }