Free cookie consent management tool by TermsFeed Policy Generator

source: branches/HeuristicLab.Crossovers/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Crossovers/SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover.cs @ 7489

Last change on this file since 7489 was 7481, checked in by mkommend, 12 years ago

#1682: Corrected gp-crossover code.

  • Changed ISymbolicExpressionTreeCrossover
  • Corrected SubtreeCrossover
  • Updated MultiSymbolicDataAnalysisCrossover
File size: 7.7 KB
Line 
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
22using System;
23using System.Collections.Generic;
24using System.Linq;
25using HeuristicLab.Common;
26using HeuristicLab.Core;
27using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
28using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
29
30namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
31
32  [Item("ProbabilisticFunctionalCrossover", "An operator which performs subtree swapping based on the behavioral similarity between subtrees.")]
33  public sealed class SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover<T> : SymbolicDataAnalysisExpressionCrossover<T> where T : class, IDataAnalysisProblemData {
34    [StorableConstructor]
35    private SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover(bool deserializing) : base(deserializing) { }
36    private SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover(SymbolicDataAnalysisExpressionCrossover<T> original, Cloner cloner)
37      : base(original, cloner) { }
38    public SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover()
39      : base() {
40    }
41    public override IDeepCloneable Clone(Cloner cloner) { return new SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover<T>(this, cloner); }
42
43    public override ISymbolicExpressionTree Crossover(IRandom random, ISymbolicExpressionTree parent0, ISymbolicExpressionTree parent1) {
44      ISymbolicDataAnalysisExpressionTreeInterpreter interpreter = SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
45      List<int> rows = GenerateRowsToEvaluate().ToList();
46      T problemData = ProblemDataParameter.ActualValue;
47      return Cross(random, parent0, parent1, interpreter, problemData,
48                   rows, MaximumSymbolicExpressionTreeDepth.Value, MaximumSymbolicExpressionTreeLength.Value);
49    }
50
51    /// <summary>
52    /// Takes two parent individuals P0 and P1.
53    /// Randomly choose a node i from the first parent, then for each matching node j from the second parent, calculate the behavioral distance based on the range:
54    /// d(i,j) = 0.5 * ( abs(max(i) - max(j)) + abs(min(i) - min(j)) ).
55    /// Next, assign probabilities for the selection of the second cross point based on the inversed and normalized behavioral distance and
56    /// choose the second crosspoint via a random weighted selection procedure.
57    /// </summary>
58    public static ISymbolicExpressionTree Cross(IRandom random, ISymbolicExpressionTree parent0, ISymbolicExpressionTree parent1,
59                                                ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, T problemData, IList<int> rows, int maxDepth, int maxLength) {
60      var crossoverPoints0 = new List<CutPoint>();
61      parent0.Root.ForEachNodePostfix((n) => {
62        if (n.Subtrees.Any() && n != parent0.Root)
63          crossoverPoints0.AddRange(from s in n.Subtrees
64                                    select new CutPoint(n, s));
65      });
66      var crossoverPoint0 = crossoverPoints0.SelectRandom(random);
67      int level = parent0.Root.GetBranchLevel(crossoverPoint0.Child);
68      int length = parent0.Root.GetLength() - crossoverPoint0.Child.GetLength();
69
70      var allowedBranches = new List<ISymbolicExpressionTreeNode>();
71      parent1.Root.ForEachNodePostfix((n) => {
72        if (n.Subtrees.Any() && n != parent1.Root)
73          allowedBranches.AddRange(from s in n.Subtrees
74                                   where crossoverPoint0.IsMatchingPointType(s) && s.GetDepth() + level <= maxDepth && s.GetLength() + length <= maxLength
75                                   select s);
76      });
77
78      if (allowedBranches.Count == 0)
79        return parent0;
80
81      var dataset = problemData.Dataset;
82
83      // create symbols in order to improvize an ad-hoc tree so that the child can be evaluated
84      var rootSymbol = new ProgramRootSymbol();
85      var startSymbol = new StartSymbol();
86      var tree0 = CreateTreeFromNode(random, crossoverPoint0.Child, rootSymbol, startSymbol); // this will change crossoverPoint0.Child.Parent
87      double min0 = 0.0, max0 = 0.0;
88      foreach (double v in interpreter.GetSymbolicExpressionTreeValues(tree0, dataset, rows)) {
89        if (min0 > v) min0 = v;
90        if (max0 < v) max0 = v;
91      }
92      crossoverPoint0.Child.Parent = crossoverPoint0.Parent; // restore correct parent
93
94      var weights = new List<double>();
95      foreach (var node in allowedBranches) {
96        var parent = node.Parent;
97        var tree1 = CreateTreeFromNode(random, node, rootSymbol, startSymbol);
98        double min1 = 0.0, max1 = 0.0;
99        foreach (double v in interpreter.GetSymbolicExpressionTreeValues(tree1, dataset, rows)) {
100          if (min1 > v) min1 = v;
101          if (max1 < v) max1 = v;
102        }
103        double behavioralDistance = (Math.Abs(min0 - min1) + Math.Abs(max0 - max1)) / 2; // this can be NaN of Infinity because some trees are crazy like exp(exp(exp(...))), we correct that below
104        weights.Add(behavioralDistance);
105        node.Parent = parent; // restore correct node parent
106      }
107
108      // remove branches with an infinite or NaN behavioral distance
109      int count = weights.Count, idx = 0;
110      while (idx < count) {
111        if (Double.IsNaN(weights[idx]) || Double.IsInfinity(weights[idx])) {
112          weights.RemoveAt(idx);
113          allowedBranches.RemoveAt(idx);
114          --count;
115        } else {
116          ++idx;
117        }
118      }
119
120      // check if there are any allowed branches left
121      if (allowedBranches.Count == 0)
122        return parent0;
123
124      ISymbolicExpressionTreeNode selectedBranch;
125      double sum = weights.Sum();
126
127      if (sum.IsAlmost(0.0) || weights.Count == 1) // if there is only one allowed branch, or if all weights are zero
128        selectedBranch = allowedBranches[0];
129      else {
130        for (int i = 0; i != weights.Count; ++i) // normalize and invert values
131          weights[i] = 1 - weights[i] / sum;
132
133        sum = weights.Sum(); // take new sum
134
135        // compute the probabilities (selection weights)
136        for (int i = 0; i != weights.Count; ++i)
137          weights[i] /= sum;
138
139        selectedBranch = allowedBranches.SelectRandom(weights, random);
140      }
141      swap(crossoverPoint0, selectedBranch);
142      return parent0;
143    }
144
145    private static void swap(CutPoint crossoverPoint, ISymbolicExpressionTreeNode selectedBranch) {
146      if (crossoverPoint.Child != null) {
147        // manipulate the tree of parent0 in place
148        // replace the branch in tree0 with the selected branch from tree1
149        crossoverPoint.Parent.RemoveSubtree(crossoverPoint.ChildIndex);
150        if (selectedBranch != null) {
151          crossoverPoint.Parent.InsertSubtree(crossoverPoint.ChildIndex, selectedBranch);
152        }
153      } else {
154        // child is null (additional child should be added under the parent)
155        if (selectedBranch != null) {
156          crossoverPoint.Parent.AddSubtree(selectedBranch);
157        }
158      }
159    }
160  }
161}
Note: See TracBrowser for help on using the repository browser.