Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Crossovers/SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover.cs @ 16565

Last change on this file since 16565 was 16565, checked in by gkronber, 5 years ago

#2520: merged changes from PersistenceOverhaul branch (r16451:16564) into trunk

File size: 7.6 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2019 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 HEAL.Attic;
29using HeuristicLab.Random;
30
31namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
32  [Item("ProbabilisticFunctionalCrossover", "An operator which performs subtree swapping based on the behavioral similarity between subtrees:\n" +
33                                            "- Take two parent individuals P0 and P1\n" +
34                                            "- Randomly choose a node N from P0\n" +
35                                            "- For each matching node M from P1, calculate the behavioral distance:\n" +
36                                            "\t\tD(N,M) = 0.5 * ( abs(max(N) - max(M)) + abs(min(N) - min(M)) )\n" +
37                                            "- Make a probabilistic weighted choice of node M from P1, based on the inversed and normalized behavioral distance")]
38  [StorableType("EC1901B8-689B-4FC9-A79E-2778B3D0609E")]
39  public sealed class SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover<T> : SymbolicDataAnalysisExpressionCrossover<T> where T : class, IDataAnalysisProblemData {
40    [StorableConstructor]
41    private SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover(StorableConstructorFlag _) : base(_) { }
42    private SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover(SymbolicDataAnalysisExpressionCrossover<T> original, Cloner cloner)
43      : base(original, cloner) { }
44    public SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover()
45      : base() {
46      name = "ProbabilisticFunctionalCrossover";
47    }
48    public override IDeepCloneable Clone(Cloner cloner) { return new SymbolicDataAnalysisExpressionProbabilisticFunctionalCrossover<T>(this, cloner); }
49
50    public override ISymbolicExpressionTree Crossover(IRandom random, ISymbolicExpressionTree parent0, ISymbolicExpressionTree parent1) {
51      ISymbolicDataAnalysisExpressionTreeInterpreter interpreter = SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
52      List<int> rows = GenerateRowsToEvaluate().ToList();
53      T problemData = ProblemDataParameter.ActualValue;
54      return Cross(random, parent0, parent1, interpreter, problemData,
55                   rows, MaximumSymbolicExpressionTreeDepth.Value, MaximumSymbolicExpressionTreeLength.Value);
56    }
57
58    /// <summary>
59    /// Takes two parent individuals P0 and P1.
60    /// 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:
61    /// d(i,j) = 0.5 * ( abs(max(i) - max(j)) + abs(min(i) - min(j)) ).
62    /// Next, assign probabilities for the selection of a node j based on the inversed and normalized behavioral distance, then make a random weighted choice.
63    /// </summary>
64    public static ISymbolicExpressionTree Cross(IRandom random, ISymbolicExpressionTree parent0, ISymbolicExpressionTree parent1,
65                                                ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, T problemData, IList<int> rows, int maxDepth, int maxLength) {
66      var crossoverPoints0 = new List<CutPoint>();
67      parent0.Root.ForEachNodePostfix((n) => {
68        // the if clauses prevent the root or the startnode from being selected, although the startnode can be the parent of the node being swapped.
69        if (n.Parent != null && n.Parent != parent0.Root) {
70          crossoverPoints0.Add(new CutPoint(n.Parent, n));
71        }
72      });
73
74      var crossoverPoint0 = crossoverPoints0.SampleRandom(random);
75      int level = parent0.Root.GetBranchLevel(crossoverPoint0.Child);
76      int length = parent0.Root.GetLength() - crossoverPoint0.Child.GetLength();
77
78      var allowedBranches = new List<ISymbolicExpressionTreeNode>();
79      parent1.Root.ForEachNodePostfix((n) => {
80        if (n.Parent != null && n.Parent != parent1.Root) {
81          if (n.GetDepth() + level <= maxDepth && n.GetLength() + length <= maxLength && crossoverPoint0.IsMatchingPointType(n))
82            allowedBranches.Add(n);
83        }
84      });
85
86      if (allowedBranches.Count == 0)
87        return parent0;
88
89      var dataset = problemData.Dataset;
90
91      // create symbols in order to improvize an ad-hoc tree so that the child can be evaluated
92      var rootSymbol = new ProgramRootSymbol();
93      var startSymbol = new StartSymbol();
94      var tree0 = CreateTreeFromNode(random, crossoverPoint0.Child, rootSymbol, startSymbol); // this will change crossoverPoint0.Child.Parent
95      double min0 = 0.0, max0 = 0.0;
96      foreach (double v in interpreter.GetSymbolicExpressionTreeValues(tree0, dataset, rows)) {
97        if (min0 > v) min0 = v;
98        if (max0 < v) max0 = v;
99      }
100      crossoverPoint0.Child.Parent = crossoverPoint0.Parent; // restore correct parent
101
102      var weights = new List<double>();
103      foreach (var node in allowedBranches) {
104        var parent = node.Parent;
105        var tree1 = CreateTreeFromNode(random, node, rootSymbol, startSymbol);
106        double min1 = 0.0, max1 = 0.0;
107        foreach (double v in interpreter.GetSymbolicExpressionTreeValues(tree1, dataset, rows)) {
108          if (min1 > v) min1 = v;
109          if (max1 < v) max1 = v;
110        }
111        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
112        weights.Add(behavioralDistance);
113        node.Parent = parent; // restore correct node parent
114      }
115
116      // remove branches with an infinite or NaN behavioral distance
117      for (int i = weights.Count - 1; i >= 0; --i) {
118        if (Double.IsNaN(weights[i]) || Double.IsInfinity(weights[i])) {
119          weights.RemoveAt(i);
120          allowedBranches.RemoveAt(i);
121        }
122      }
123      // check if there are any allowed branches left
124      if (allowedBranches.Count == 0)
125        return parent0;
126
127      ISymbolicExpressionTreeNode selectedBranch;
128      double sum = weights.Sum();
129
130      if (sum.IsAlmost(0.0) || weights.Count == 1) // if there is only one allowed branch, or if all weights are zero
131        selectedBranch = allowedBranches[0];
132      else {
133        for (int i = 0; i != weights.Count; ++i) // normalize and invert values
134          weights[i] = 1 - weights[i] / sum;
135
136        sum = weights.Sum(); // take new sum
137
138        // compute the probabilities (selection weights)
139        for (int i = 0; i != weights.Count; ++i)
140          weights[i] /= sum;
141
142#pragma warning disable 612, 618
143        selectedBranch = allowedBranches.SelectRandom(weights, random);
144#pragma warning restore 612, 618
145      }
146      Swap(crossoverPoint0, selectedBranch);
147      return parent0;
148    }
149  }
150}
Note: See TracBrowser for help on using the repository browser.