Free cookie consent management tool by TermsFeed Policy Generator

source: branches/1837_Sliding Window GP/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer.cs @ 17167

Last change on this file since 17167 was 9708, checked in by mkommend, 11 years ago

#1837: Merged trunk changes into sliding window gp branch.

File size: 10.4 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2013 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.Data;
28using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
29using HeuristicLab.Optimization;
30using HeuristicLab.Parameters;
31using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
32
33namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
34  /// <summary>
35  /// An operator that collects the Pareto-best symbolic data analysis solutions for single objective symbolic data analysis problems.
36  /// </summary>
37  [Item("SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer", "An operator that analyzes the Pareto-best symbolic data analysis solution for single objective symbolic data analysis problems.")]
38  [StorableClass]
39  public abstract class SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer<S, T> : SymbolicDataAnalysisSingleObjectiveAnalyzer, ISymbolicDataAnalysisInterpreterOperator, ISymbolicDataAnalysisBoundedOperator
40    where T : class, ISymbolicDataAnalysisSolution
41    where S : class, IDataAnalysisProblemData {
42    private const string ProblemDataParameterName = "ProblemData";
43    private const string TrainingBestSolutionsParameterName = "Best training solutions";
44    private const string TrainingBestSolutionQualitiesParameterName = "Best training solution qualities";
45    private const string ComplexityParameterName = "Complexity";
46    private const string SymbolicDataAnalysisTreeInterpreterParameterName = "SymbolicDataAnalysisTreeInterpreter";
47    private const string EstimationLimitsParameterName = "EstimationLimits";
48
49    public override bool EnabledByDefault {
50      get { return false; }
51    }
52
53    #region parameter properties
54    public ILookupParameter<ItemList<T>> TrainingBestSolutionsParameter {
55      get { return (ILookupParameter<ItemList<T>>)Parameters[TrainingBestSolutionsParameterName]; }
56    }
57    public ILookupParameter<ItemList<DoubleArray>> TrainingBestSolutionQualitiesParameter {
58      get { return (ILookupParameter<ItemList<DoubleArray>>)Parameters[TrainingBestSolutionQualitiesParameterName]; }
59    }
60    public IScopeTreeLookupParameter<DoubleValue> ComplexityParameter {
61      get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters[ComplexityParameterName]; }
62    }
63    public ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter> SymbolicDataAnalysisTreeInterpreterParameter {
64      get { return (ILookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>)Parameters[SymbolicDataAnalysisTreeInterpreterParameterName]; }
65    }
66    public ILookupParameter<S> ProblemDataParameter {
67      get { return (ILookupParameter<S>)Parameters[ProblemDataParameterName]; }
68    }
69    public IValueLookupParameter<DoubleLimit> EstimationLimitsParameter {
70      get { return (IValueLookupParameter<DoubleLimit>)Parameters[EstimationLimitsParameterName]; }
71    }
72    #endregion
73    #region properties
74    public ItemList<T> TrainingBestSolutions {
75      get { return TrainingBestSolutionsParameter.ActualValue; }
76      set { TrainingBestSolutionsParameter.ActualValue = value; }
77    }
78    public ItemList<DoubleArray> TrainingBestSolutionQualities {
79      get { return TrainingBestSolutionQualitiesParameter.ActualValue; }
80      set { TrainingBestSolutionQualitiesParameter.ActualValue = value; }
81    }
82    #endregion
83
84    [StorableConstructor]
85    protected SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer(bool deserializing) : base(deserializing) { }
86    protected SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer(SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer<S, T> original, Cloner cloner) : base(original, cloner) { }
87    public SymbolicDataAnalysisSingleObjectiveTrainingParetoBestSolutionAnalyzer()
88      : base() {
89      Parameters.Add(new LookupParameter<S>(ProblemDataParameterName, "The problem data for the symbolic data analysis solution."));
90      Parameters.Add(new LookupParameter<ItemList<T>>(TrainingBestSolutionsParameterName, "The training best (Pareto-optimal) symbolic data analysis solutions."));
91      Parameters.Add(new LookupParameter<ItemList<DoubleArray>>(TrainingBestSolutionQualitiesParameterName, "The qualities of the training best (Pareto-optimal) solutions."));
92      Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>(ComplexityParameterName, "The complexity of each tree."));
93      Parameters.Add(new LookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>(SymbolicDataAnalysisTreeInterpreterParameterName, "The symbolic data analysis tree interpreter for the symbolic expression tree."));
94      Parameters.Add(new ValueLookupParameter<DoubleLimit>(EstimationLimitsParameterName, "The lower and upper limit for the estimated values produced by the symbolic classification model."));
95    }
96
97    public override IOperation Apply() {
98      var results = ResultCollection;
99      // create empty parameter and result values
100      if (TrainingBestSolutions == null) {
101        TrainingBestSolutions = new ItemList<T>();
102        TrainingBestSolutionQualities = new ItemList<DoubleArray>();
103        results.Add(new Result(TrainingBestSolutionQualitiesParameter.Name, TrainingBestSolutionQualitiesParameter.Description, TrainingBestSolutionQualities));
104        results.Add(new Result(TrainingBestSolutionsParameter.Name, TrainingBestSolutionsParameter.Description, TrainingBestSolutions));
105      }
106
107      IList<Tuple<double, double>> trainingBestQualities = TrainingBestSolutionQualities
108        .Select(x => Tuple.Create(x[0], x[1]))
109        .ToList();
110
111      #region find best trees
112      IList<int> nonDominatedIndexes = new List<int>();
113      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
114      List<double> qualities = Quality.Select(x => x.Value).ToList();
115
116      List<double> complexities;
117      if (ComplexityParameter.ActualValue != null && ComplexityParameter.ActualValue.Length == qualities.Count) {
118        complexities = ComplexityParameter.ActualValue.Select(x => x.Value).ToList();
119      } else {
120        complexities = tree.Select(t => (double)t.Length).ToList();
121      }
122      List<Tuple<double, double>> fitness = new List<Tuple<double, double>>();
123      for (int i = 0; i < qualities.Count; i++)
124        fitness.Add(Tuple.Create(qualities[i], complexities[i]));
125      var maximization = Tuple.Create(Maximization.Value, false);// complexity must be minimized
126      List<Tuple<double, double>> newNonDominatedQualities = new List<Tuple<double, double>>();
127      for (int i = 0; i < tree.Length; i++) {
128        if (IsNonDominated(fitness[i], trainingBestQualities, maximization) &&
129          IsNonDominated(fitness[i], newNonDominatedQualities, maximization) &&
130          IsNonDominated(fitness[i], fitness.Skip(i + 1), maximization)) {
131          if (!newNonDominatedQualities.Contains(fitness[i])) {
132            newNonDominatedQualities.Add(fitness[i]);
133            nonDominatedIndexes.Add(i);
134          }
135        }
136      }
137      #endregion
138
139      #region update Pareto-optimal solution archive
140      if (nonDominatedIndexes.Count > 0) {
141        ItemList<DoubleArray> nonDominatedQualities = new ItemList<DoubleArray>();
142        ItemList<T> nonDominatedSolutions = new ItemList<T>();
143        // add all new non-dominated solutions to the archive
144        foreach (var index in nonDominatedIndexes) {
145          T solution = CreateSolution(tree[index]);
146          nonDominatedSolutions.Add(solution);
147          nonDominatedQualities.Add(new DoubleArray(new double[] { fitness[index].Item1, fitness[index].Item2 }));
148        }
149        // add old non-dominated solutions only if they are not dominated by one of the new solutions
150        for (int i = 0; i < trainingBestQualities.Count; i++) {
151          if (IsNonDominated(trainingBestQualities[i], newNonDominatedQualities, maximization)) {
152            if (!newNonDominatedQualities.Contains(trainingBestQualities[i])) {
153              nonDominatedSolutions.Add(TrainingBestSolutions[i]);
154              nonDominatedQualities.Add(TrainingBestSolutionQualities[i]);
155            }
156          }
157        }
158
159        // make sure solutions and qualities are ordered in the results
160        var orderedIndexes =
161          nonDominatedSolutions.Select((s, i) => i).OrderBy(i => nonDominatedQualities[i][0]).ToArray();
162
163        var orderedNonDominatedSolutions = new ItemList<T>();
164        var orderedNonDominatedQualities = new ItemList<DoubleArray>();
165        foreach (var i in orderedIndexes) {
166          orderedNonDominatedQualities.Add(nonDominatedQualities[i]);
167          orderedNonDominatedSolutions.Add(nonDominatedSolutions[i]);
168        }
169
170        TrainingBestSolutions = orderedNonDominatedSolutions;
171        TrainingBestSolutionQualities = orderedNonDominatedQualities;
172
173        results[TrainingBestSolutionsParameter.Name].Value = orderedNonDominatedSolutions;
174        results[TrainingBestSolutionQualitiesParameter.Name].Value = orderedNonDominatedQualities;
175      }
176      #endregion
177      return base.Apply();
178    }
179
180    protected abstract T CreateSolution(ISymbolicExpressionTree bestTree);
181
182    private bool IsNonDominated(Tuple<double, double> point, IEnumerable<Tuple<double, double>> points, Tuple<bool, bool> maximization) {
183      return !points.Any(p => IsBetterOrEqual(p.Item1, point.Item1, maximization.Item1) &&
184                             IsBetterOrEqual(p.Item2, point.Item2, maximization.Item2));
185    }
186    private bool IsBetterOrEqual(double lhs, double rhs, bool maximization) {
187      if (maximization) return lhs >= rhs;
188      else return lhs <= rhs;
189    }
190  }
191}
Note: See TracBrowser for help on using the repository browser.