Changeset 8422

Timestamp:

08/07/12 10:40:39 (12 years ago)

Author:

abeham

Message:

#1913:

• Reduced memory footprint
• Fixed some bugs
• Added some tests

Location:

branches/NCA

Files:

• . (modified) (1 prop)
• HeuristicLab.Algorithms.NCA.Tests (added)
• HeuristicLab.Algorithms.NCA.Tests/HeuristicLab.Algorithms.NCA.Tests.csproj (added)
• HeuristicLab.Algorithms.NCA.Tests/MatrixTest.cs (added)
• HeuristicLab.Algorithms.NCA.Tests/Properties (added)
• HeuristicLab.Algorithms.NCA.Tests/Properties/AssemblyInfo.cs (added)
• HeuristicLab.Algorithms.NCA.Tests/Test References (added)
• HeuristicLab.Algorithms.NCA.Tests/Test References/HeuristicLab.Algorithms.NCA-3.3.accessor (added)
• HeuristicLab.Algorithms.NCA/3.3/Matrix.cs (modified) (1 diff)
• HeuristicLab.Algorithms.NCA/3.3/NeighborhoodComponentsAnalysis.cs (modified) (4 diffs)
• Local.testsettings (added)
• NCA.sln (modified) (2 diffs)
• NCA.vsmdi (added)
• TraceAndTestImpact.testsettings (added)

branches/NCA

@@ -1 +1 @@
 *.suo
+*.user
+TestResults

@@ -1 +1 @@
 *.suo
+*.user
+TestResults

branches/NCA/HeuristicLab.Algorithms.NCA/3.3/Matrix.cs

@@ -122 +122 @@
     }
 
-    public double QuadraticSum() {
+    public double Length() {
       if (Rows != 1) throw new ArgumentException("Length only works on vectors.");
-      return values.Sum(x => x * x);
+      return Math.Sqrt(values.Sum(x => x * x));
     }
 

branches/NCA/HeuristicLab.Algorithms.NCA/3.3/NeighborhoodComponentsAnalysis.cs

@@ -32 +32 @@
       var instances = data.TrainingIndices.Count();
       var attributes = data.AllowedInputVariables.Count();
+
+      /*alglib.pcabuildbasis(
+    double[,] x,
+    int npoints,
+    int nvars,
+    out int info,
+    out double[] s2,
+    out double[,] v)*/
+
       var matrix = new double[attributes * reduceDimensions];
+
       // TODO: make some more clever initialization of matrix
       var rand = new Random();
@@ -58 +68 @@
       var rowCount = 0;
       foreach (var r in data.TrainingIndices) {
-        var j = 0;
+        var i = 0;
         foreach (var v in data.AllowedInputVariables) {
           var val = data.Dataset.GetDoubleValue(v, r);
-          for (var i = 0; i < transformedTrainingset.GetLength(1); i++)
-            transformedTrainingset[rowCount, i] += val * transformationMatrix[j, i];
-          j++;
+          for (var j = 0; j < reduceDimensions; j++)
+            transformedTrainingset[rowCount, j] += val * transformationMatrix[i, j];
+          i++;
         }
         rowCount++;
@@ -79 +89 @@
       var info = (OptimizationInfo)obj;
       var instances = info.ProblemData.TrainingIndices.ToArray();
+      var attributes = info.ProblemData.AllowedInputVariables.Count();
       var AMatrix = new Matrix(A, A.Length / info.ReduceDimensions, info.ReduceDimensions);
-      var distances = new double[instances.Length, instances.Length];
 
+      alglib.sparsematrix probabilities;
+      alglib.sparsecreate(instances.Length, instances.Length, out probabilities);
+      var distances = new double[instances.Length];
       for (int i = 0; i < instances.Length - 1; i++) {
         var iVector = new Matrix(GetRow(info.ProblemData, instances[i]));
-        for (int j = i + 1; j < instances.Length; j++) {
-          var jVector = new Matrix(GetRow(info.ProblemData, instances[j]));
-          distances[i, j] = iVector.Multiply(AMatrix).Subtract(jVector.Multiply(AMatrix)).QuadraticSum();
-          distances[j, i] = distances[i, j];
-        }
-      }
-      alglib.sparsematrix probabilities;
-      alglib.sparsecreate(instances.Length, instances.Length, out probabilities);
-      for (int i = 0; i < instances.Length; i++) {
         var denom = 0.0;
         for (int k = 0; k < instances.Length; k++) {
           if (k == i) continue;
-          denom += Math.Exp(-(distances[i, k] * distances[i, k]));
+          var kVector = new Matrix(GetRow(info.ProblemData, instances[k]));
+          distances[k] = iVector.Multiply(AMatrix).Subtract(kVector.Multiply(AMatrix)).Length();
+          denom += Math.Exp(-(distances[k] * distances[k]));
         }
         if (denom > 0) {
-          for (int j = 0; j < instances.Length; j++) {
+          for (int j = i + 1; j < instances.Length; j++) {
             if (i == j) continue;
-            var v = Math.Exp(-(distances[i, j] * distances[i, j])) / denom;
+            var v = Math.Exp(-(distances[j] * distances[j])) / denom;
             alglib.sparseset(probabilities, i, j, v);
+            alglib.sparseset(probabilities, j, i, v);
           }
         }
       }
-      alglib.sparseconverttocrs(probabilities); // needed to enumerate in order
+      alglib.sparseconverttocrs(probabilities); // needed to enumerate in order (top-down and left-right)
 
       int t0 = 0, t1 = 0, r, c;
@@ -116 +123 @@
           pi[r] += val;
       }
-      func = pi.Sum();
 
-      grad = AMatrix.Multiply(2.0).Transpose().Multiply(CalculateInnerSum(probabilities, pi, info.ProblemData, instances, classes)).Transpose().ToArray();
-    }
-
-    private static Matrix CalculateInnerSum(alglib.sparsematrix p, double[] pi, IClassificationProblemData data, int[] instances, double[] classes) {
-      var attributes = data.AllowedInputVariables.Count();
-      var target = data.TargetVariable;
-      int t0 = 0, t1 = 0, r, c;
-      double v;
-      var result = new double[attributes, attributes];
-      while (alglib.sparseenumerate(p, ref t0, ref t1, out r, out c, out v)) {
-        var vector = new Matrix(GetRow(data, instances[r])).Subtract(new Matrix(GetRow(data, instances[c]))).Apply();
-        vector.OuterProduct(vector).Multiply(v * pi[r]).AddTo(result);
+      var innerSum = new double[attributes, attributes];
+      while (alglib.sparseenumerate(probabilities, ref t0, ref t1, out r, out c, out val)) {
+        var vector = new Matrix(GetRow(info.ProblemData, instances[r])).Subtract(new Matrix(GetRow(info.ProblemData, instances[c]))).Apply();
+        vector.OuterProduct(vector).Multiply(val * pi[r]).AddTo(innerSum);
 
         if (classes[r].IsAlmost(classes[c])) {
-          vector.OuterProduct(vector).Multiply(-v).AddTo(result);
+          vector.OuterProduct(vector).Multiply(-val).AddTo(innerSum);
         }
       }
-      return new Matrix(result);
+
+      func = -pi.Sum();
+
+      grad = AMatrix.Multiply(-2.0).Transpose().Multiply(new Matrix(innerSum)).Transpose().ToArray();
     }
 

branches/NCA/NCA.sln

@@ -8 +8 @@
 Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{30CDFA65-3AD8-42A4-801E-0F785CE9DF16}"
   ProjectSection(SolutionItems) = preProject
+    Local.testsettings = Local.testsettings
+    NCA.vsmdi = NCA.vsmdi
     PreBuildEvent.cmd = PreBuildEvent.cmd
+    TraceAndTestImpact.testsettings = TraceAndTestImpact.testsettings
   EndProjectSection
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "HeuristicLab.Algorithms.NCA.Tests", "HeuristicLab.Algorithms.NCA.Tests\HeuristicLab.Algorithms.NCA.Tests.csproj", "{910C130C-378D-422B-96B7-0484F7CC66B7}"
+EndProject
 Global
+  GlobalSection(TestCaseManagementSettings) = postSolution
+    CategoryFile = NCA.vsmdi
+  EndGlobalSection
   GlobalSection(SolutionConfigurationPlatforms) = preSolution
     Debug|Any CPU = Debug|Any CPU
@@ -41 +49 @@
     {07654A26-5964-4079-B023-5548B1EB1D1E}.Release|x86.ActiveCfg = Release|x86
     {07654A26-5964-4079-B023-5548B1EB1D1E}.Release|x86.Build.0 = Release|x86
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Debug|Any CPU.Build.0 = Debug|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Debug|x64.ActiveCfg = Debug|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Debug|x86.ActiveCfg = Debug|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Release|Any CPU.ActiveCfg = Release|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Release|Any CPU.Build.0 = Release|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Release|x64.ActiveCfg = Release|Any CPU
+    {910C130C-378D-422B-96B7-0484F7CC66B7}.Release|x86.ActiveCfg = Release|Any CPU
   EndGlobalSection
   GlobalSection(SolutionProperties) = preSolution

@@ -1 +1 @@
 *.suo
+*.user
+TestResults