1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2014 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 |
|
---|
22 | using System.Collections.Generic;
|
---|
23 | using System.Linq;
|
---|
24 | using HeuristicLab.Common;
|
---|
25 | using HeuristicLab.Core;
|
---|
26 | using HeuristicLab.Data;
|
---|
27 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
28 | using HeuristicLab.Optimization;
|
---|
29 | using HeuristicLab.Parameters;
|
---|
30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
31 |
|
---|
32 | namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
|
---|
33 | /// <summary>
|
---|
34 | /// An operator that analyzes the training best symbolic data analysis solution for multi objective symbolic data analysis problems.
|
---|
35 | /// </summary>
|
---|
36 | [Item("SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer", "An operator that analyzes the training best symbolic data analysis solution for multi objective symbolic data analysis problems.")]
|
---|
37 | [StorableClass]
|
---|
38 | public abstract class SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer<T> : SymbolicDataAnalysisMultiObjectiveAnalyzer
|
---|
39 | where T : class, ISymbolicDataAnalysisSolution {
|
---|
40 | private const string TrainingBestSolutionsParameterName = "Best training solutions";
|
---|
41 | private const string TrainingBestSolutionQualitiesParameterName = "Best training solution qualities";
|
---|
42 | private const string UpdateAlwaysParameterName = "Always update best solutions";
|
---|
43 |
|
---|
44 | private const string TrainingBestSolutionParameterName = "Best training solution";
|
---|
45 | private const string TrainingBestSolutionQualityParameterName = "Best training solution quality";
|
---|
46 | private const string TrainingBestSolutionGenerationParameterName = "Best training solution generation";
|
---|
47 |
|
---|
48 | #region parameter properties
|
---|
49 | public ILookupParameter<ItemList<T>> TrainingBestSolutionsParameter {
|
---|
50 | get { return (ILookupParameter<ItemList<T>>)Parameters[TrainingBestSolutionsParameterName]; }
|
---|
51 | }
|
---|
52 | public ILookupParameter<ItemList<DoubleArray>> TrainingBestSolutionQualitiesParameter {
|
---|
53 | get { return (ILookupParameter<ItemList<DoubleArray>>)Parameters[TrainingBestSolutionQualitiesParameterName]; }
|
---|
54 | }
|
---|
55 | public IFixedValueParameter<BoolValue> UpdateAlwaysParameter {
|
---|
56 | get { return (IFixedValueParameter<BoolValue>)Parameters[UpdateAlwaysParameterName]; }
|
---|
57 | }
|
---|
58 | #endregion
|
---|
59 | #region properties
|
---|
60 | public ItemList<T> TrainingBestSolutions {
|
---|
61 | get { return TrainingBestSolutionsParameter.ActualValue; }
|
---|
62 | set { TrainingBestSolutionsParameter.ActualValue = value; }
|
---|
63 | }
|
---|
64 | public ItemList<DoubleArray> TrainingBestSolutionQualities {
|
---|
65 | get { return TrainingBestSolutionQualitiesParameter.ActualValue; }
|
---|
66 | set { TrainingBestSolutionQualitiesParameter.ActualValue = value; }
|
---|
67 | }
|
---|
68 | public BoolValue UpdateAlways {
|
---|
69 | get { return UpdateAlwaysParameter.Value; }
|
---|
70 | }
|
---|
71 | #endregion
|
---|
72 |
|
---|
73 | [StorableConstructor]
|
---|
74 | protected SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer(bool deserializing) : base(deserializing) { }
|
---|
75 | protected SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer(SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer<T> original, Cloner cloner) : base(original, cloner) { }
|
---|
76 | public SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer()
|
---|
77 | : base() {
|
---|
78 | Parameters.Add(new LookupParameter<ItemList<T>>(TrainingBestSolutionsParameterName, "The training best (Pareto-optimal) symbolic data analysis solutions."));
|
---|
79 | Parameters.Add(new LookupParameter<ItemList<DoubleArray>>(TrainingBestSolutionQualitiesParameterName, "The qualities of the training best (Pareto-optimal) solutions."));
|
---|
80 | Parameters.Add(new FixedValueParameter<BoolValue>(UpdateAlwaysParameterName, "Determines if the best training solutions should always be updated regardless of its quality.", new BoolValue(false)));
|
---|
81 | UpdateAlwaysParameter.Hidden = true;
|
---|
82 | }
|
---|
83 |
|
---|
84 | [StorableHook(HookType.AfterDeserialization)]
|
---|
85 | private void AfterDeserialization() {
|
---|
86 | if (!Parameters.ContainsKey(UpdateAlwaysParameterName)) {
|
---|
87 | Parameters.Add(new FixedValueParameter<BoolValue>(UpdateAlwaysParameterName, "Determines if the best training solutions should always be updated regardless of its quality.", new BoolValue(false)));
|
---|
88 | UpdateAlwaysParameter.Hidden = true;
|
---|
89 | }
|
---|
90 | }
|
---|
91 |
|
---|
92 | public override IOperation Apply() {
|
---|
93 | var results = ResultCollection;
|
---|
94 | // create empty parameter and result values
|
---|
95 | if (TrainingBestSolutions == null) {
|
---|
96 | TrainingBestSolutions = new ItemList<T>();
|
---|
97 | TrainingBestSolutionQualities = new ItemList<DoubleArray>();
|
---|
98 | results.Add(new Result(TrainingBestSolutionQualitiesParameter.Name, TrainingBestSolutionQualitiesParameter.Description, TrainingBestSolutionQualities));
|
---|
99 | results.Add(new Result(TrainingBestSolutionsParameter.Name, TrainingBestSolutionsParameter.Description, TrainingBestSolutions));
|
---|
100 | }
|
---|
101 |
|
---|
102 | if (!results.ContainsKey(TrainingBestSolutionParameterName)) {
|
---|
103 | results.Add(new Result(TrainingBestSolutionParameterName, "", typeof(ISymbolicDataAnalysisSolution)));
|
---|
104 | }
|
---|
105 |
|
---|
106 | //if the pareto front of best solutions shall be updated regardless of the quality, the list initialized empty to discard old solutions
|
---|
107 | List<double[]> trainingBestQualities;
|
---|
108 | if (UpdateAlways.Value) {
|
---|
109 | trainingBestQualities = new List<double[]>();
|
---|
110 | } else {
|
---|
111 | trainingBestQualities = TrainingBestSolutionQualities.Select(x => x.ToArray()).ToList();
|
---|
112 | }
|
---|
113 |
|
---|
114 | ISymbolicExpressionTree[] trees = SymbolicExpressionTree.ToArray();
|
---|
115 | List<double[]> qualities = Qualities.Select(x => x.ToArray()).ToList();
|
---|
116 | bool[] maximization = Maximization.ToArray();
|
---|
117 |
|
---|
118 | var nonDominatedInvididuals = new[] { new { Tree = default(ISymbolicExpressionTree), Qualities = default(double[]) } }.ToList();
|
---|
119 | nonDominatedInvididuals.Clear();
|
---|
120 |
|
---|
121 | // build list of new non-dominated solutions
|
---|
122 | for (int i = 0; i < trees.Length; i++) {
|
---|
123 | if (IsNonDominated(qualities[i], nonDominatedInvididuals.Select(ind => ind.Qualities), maximization) &&
|
---|
124 | IsNonDominated(qualities[i], trainingBestQualities, maximization)) {
|
---|
125 | for (int j = nonDominatedInvididuals.Count - 1; j >= 0; j--) {
|
---|
126 | if (IsBetterOrEqual(qualities[i], nonDominatedInvididuals[j].Qualities, maximization)) {
|
---|
127 | nonDominatedInvididuals.RemoveAt(j);
|
---|
128 | }
|
---|
129 | }
|
---|
130 | nonDominatedInvididuals.Add(new { Tree = trees[i], Qualities = qualities[i] });
|
---|
131 | }
|
---|
132 | }
|
---|
133 |
|
---|
134 | var nonDominatedSolutions = nonDominatedInvididuals.Select(x => new { Solution = CreateSolution(x.Tree, x.Qualities), Qualities = x.Qualities }).ToList();
|
---|
135 | nonDominatedSolutions.ForEach(s => s.Solution.Name = string.Join(",", s.Qualities.Select(q => q.ToString())));
|
---|
136 |
|
---|
137 | #region update Pareto-optimal solution archive
|
---|
138 | if (nonDominatedSolutions.Count > 0) {
|
---|
139 | //add old non-dominated solutions only if they are not dominated by one of the new solutions
|
---|
140 | for (int i = 0; i < trainingBestQualities.Count; i++) {
|
---|
141 | if (IsNonDominated(trainingBestQualities[i], nonDominatedSolutions.Select(x => x.Qualities), maximization)) {
|
---|
142 | nonDominatedSolutions.Add(new { Solution = TrainingBestSolutions[i], Qualities = TrainingBestSolutionQualities[i].ToArray() });
|
---|
143 | }
|
---|
144 | }
|
---|
145 |
|
---|
146 | var sortedNonDominatedSolutions = nonDominatedSolutions.OrderByDescending(x => x.Qualities[0]);
|
---|
147 | var trainingBestSolution = sortedNonDominatedSolutions.Select(s => s.Solution).First();
|
---|
148 | results[TrainingBestSolutionParameterName].Value = trainingBestSolution;
|
---|
149 | TrainingBestSolutions = new ItemList<T>(sortedNonDominatedSolutions.Select(x => x.Solution));
|
---|
150 | results[TrainingBestSolutionsParameter.Name].Value = TrainingBestSolutions;
|
---|
151 | TrainingBestSolutionQualities = new ItemList<DoubleArray>(sortedNonDominatedSolutions.Select(x => new DoubleArray(x.Qualities)));
|
---|
152 | results[TrainingBestSolutionQualitiesParameter.Name].Value = TrainingBestSolutionQualities;
|
---|
153 | }
|
---|
154 | #endregion
|
---|
155 | return base.Apply();
|
---|
156 | }
|
---|
157 |
|
---|
158 | protected abstract T CreateSolution(ISymbolicExpressionTree bestTree, double[] bestQuality);
|
---|
159 |
|
---|
160 | private bool IsNonDominated(double[] point, IEnumerable<double[]> points, bool[] maximization) {
|
---|
161 | foreach (var refPoint in points) {
|
---|
162 | bool refPointDominatesPoint = IsBetterOrEqual(refPoint, point, maximization);
|
---|
163 | if (refPointDominatesPoint) return false;
|
---|
164 | }
|
---|
165 | return true;
|
---|
166 | }
|
---|
167 |
|
---|
168 | private bool IsBetterOrEqual(double[] lhs, double[] rhs, bool[] maximization) {
|
---|
169 | for (int i = 0; i < lhs.Length; i++) {
|
---|
170 | var result = IsBetterOrEqual(lhs[i], rhs[i], maximization[i]);
|
---|
171 | if (!result) return false;
|
---|
172 | }
|
---|
173 | return true;
|
---|
174 | }
|
---|
175 |
|
---|
176 | private bool IsBetterOrEqual(double lhs, double rhs, bool maximization) {
|
---|
177 | if (maximization) return lhs >= rhs;
|
---|
178 | else return lhs <= rhs;
|
---|
179 | }
|
---|
180 | }
|
---|
181 | }
|
---|