[5557] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[14185] | 3 | * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[5557] | 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
|
---|
[5607] | 39 | where T : class, ISymbolicDataAnalysisSolution {
|
---|
[5557] | 40 | private const string TrainingBestSolutionsParameterName = "Best training solutions";
|
---|
| 41 | private const string TrainingBestSolutionQualitiesParameterName = "Best training solution qualities";
|
---|
[9152] | 42 | private const string UpdateAlwaysParameterName = "Always update best solutions";
|
---|
[13241] | 43 | private const string TrainingBestSolutionParameterName = "Best training solution";
|
---|
[5557] | 44 |
|
---|
| 45 | #region parameter properties
|
---|
| 46 | public ILookupParameter<ItemList<T>> TrainingBestSolutionsParameter {
|
---|
| 47 | get { return (ILookupParameter<ItemList<T>>)Parameters[TrainingBestSolutionsParameterName]; }
|
---|
| 48 | }
|
---|
| 49 | public ILookupParameter<ItemList<DoubleArray>> TrainingBestSolutionQualitiesParameter {
|
---|
| 50 | get { return (ILookupParameter<ItemList<DoubleArray>>)Parameters[TrainingBestSolutionQualitiesParameterName]; }
|
---|
| 51 | }
|
---|
[9152] | 52 | public IFixedValueParameter<BoolValue> UpdateAlwaysParameter {
|
---|
| 53 | get { return (IFixedValueParameter<BoolValue>)Parameters[UpdateAlwaysParameterName]; }
|
---|
| 54 | }
|
---|
[5557] | 55 | #endregion
|
---|
| 56 | #region properties
|
---|
[13241] | 57 | private ItemList<T> TrainingBestSolutions {
|
---|
[5557] | 58 | get { return TrainingBestSolutionsParameter.ActualValue; }
|
---|
| 59 | set { TrainingBestSolutionsParameter.ActualValue = value; }
|
---|
| 60 | }
|
---|
[13241] | 61 | private ItemList<DoubleArray> TrainingBestSolutionQualities {
|
---|
[5557] | 62 | get { return TrainingBestSolutionQualitiesParameter.ActualValue; }
|
---|
| 63 | set { TrainingBestSolutionQualitiesParameter.ActualValue = value; }
|
---|
| 64 | }
|
---|
[13241] | 65 | public bool UpdateAlways {
|
---|
| 66 | get { return UpdateAlwaysParameter.Value.Value; }
|
---|
| 67 | set { UpdateAlwaysParameter.Value.Value = value; }
|
---|
[9152] | 68 | }
|
---|
[5557] | 69 | #endregion
|
---|
| 70 |
|
---|
| 71 | [StorableConstructor]
|
---|
| 72 | protected SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer(bool deserializing) : base(deserializing) { }
|
---|
| 73 | protected SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer(SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer<T> original, Cloner cloner) : base(original, cloner) { }
|
---|
| 74 | public SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer()
|
---|
| 75 | : base() {
|
---|
[5607] | 76 | Parameters.Add(new LookupParameter<ItemList<T>>(TrainingBestSolutionsParameterName, "The training best (Pareto-optimal) symbolic data analysis solutions."));
|
---|
[5557] | 77 | Parameters.Add(new LookupParameter<ItemList<DoubleArray>>(TrainingBestSolutionQualitiesParameterName, "The qualities of the training best (Pareto-optimal) solutions."));
|
---|
[9152] | 78 | Parameters.Add(new FixedValueParameter<BoolValue>(UpdateAlwaysParameterName, "Determines if the best training solutions should always be updated regardless of its quality.", new BoolValue(false)));
|
---|
| 79 | UpdateAlwaysParameter.Hidden = true;
|
---|
[5557] | 80 | }
|
---|
| 81 |
|
---|
[9152] | 82 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 83 | private void AfterDeserialization() {
|
---|
| 84 | if (!Parameters.ContainsKey(UpdateAlwaysParameterName)) {
|
---|
| 85 | Parameters.Add(new FixedValueParameter<BoolValue>(UpdateAlwaysParameterName, "Determines if the best training solutions should always be updated regardless of its quality.", new BoolValue(false)));
|
---|
| 86 | UpdateAlwaysParameter.Hidden = true;
|
---|
| 87 | }
|
---|
| 88 | }
|
---|
| 89 |
|
---|
[5557] | 90 | public override IOperation Apply() {
|
---|
| 91 | var results = ResultCollection;
|
---|
| 92 | // create empty parameter and result values
|
---|
| 93 | if (TrainingBestSolutions == null) {
|
---|
| 94 | TrainingBestSolutions = new ItemList<T>();
|
---|
| 95 | TrainingBestSolutionQualities = new ItemList<DoubleArray>();
|
---|
[5747] | 96 | results.Add(new Result(TrainingBestSolutionQualitiesParameter.Name, TrainingBestSolutionQualitiesParameter.Description, TrainingBestSolutionQualities));
|
---|
| 97 | results.Add(new Result(TrainingBestSolutionsParameter.Name, TrainingBestSolutionsParameter.Description, TrainingBestSolutions));
|
---|
[5557] | 98 | }
|
---|
| 99 |
|
---|
[13241] | 100 | if (!results.ContainsKey(TrainingBestSolutionParameterName)) {
|
---|
| 101 | results.Add(new Result(TrainingBestSolutionParameterName, "", typeof(ISymbolicDataAnalysisSolution)));
|
---|
| 102 | }
|
---|
| 103 |
|
---|
[9152] | 104 | //if the pareto front of best solutions shall be updated regardless of the quality, the list initialized empty to discard old solutions
|
---|
[13241] | 105 | List<double[]> trainingBestQualities;
|
---|
| 106 | if (UpdateAlways) {
|
---|
[9152] | 107 | trainingBestQualities = new List<double[]>();
|
---|
| 108 | } else {
|
---|
| 109 | trainingBestQualities = TrainingBestSolutionQualities.Select(x => x.ToArray()).ToList();
|
---|
| 110 | }
|
---|
[5557] | 111 |
|
---|
[13241] | 112 | ISymbolicExpressionTree[] trees = SymbolicExpressionTree.ToArray();
|
---|
[5557] | 113 | List<double[]> qualities = Qualities.Select(x => x.ToArray()).ToList();
|
---|
| 114 | bool[] maximization = Maximization.ToArray();
|
---|
[13241] | 115 |
|
---|
[13307] | 116 | var nonDominatedIndividuals = new[] { new { Tree = default(ISymbolicExpressionTree), Qualities = default(double[]) } }.ToList();
|
---|
| 117 | nonDominatedIndividuals.Clear();
|
---|
[13241] | 118 |
|
---|
| 119 | // build list of new non-dominated solutions
|
---|
| 120 | for (int i = 0; i < trees.Length; i++) {
|
---|
[13307] | 121 | if (IsNonDominated(qualities[i], nonDominatedIndividuals.Select(ind => ind.Qualities), maximization) &&
|
---|
[13241] | 122 | IsNonDominated(qualities[i], trainingBestQualities, maximization)) {
|
---|
[13307] | 123 | for (int j = nonDominatedIndividuals.Count - 1; j >= 0; j--) {
|
---|
| 124 | if (IsBetterOrEqual(qualities[i], nonDominatedIndividuals[j].Qualities, maximization)) {
|
---|
| 125 | nonDominatedIndividuals.RemoveAt(j);
|
---|
[13241] | 126 | }
|
---|
[5742] | 127 | }
|
---|
[13307] | 128 | nonDominatedIndividuals.Add(new { Tree = trees[i], Qualities = qualities[i] });
|
---|
[5557] | 129 | }
|
---|
| 130 | }
|
---|
[13241] | 131 |
|
---|
[13307] | 132 | var nonDominatedSolutions = nonDominatedIndividuals.Select(x => new { Solution = CreateSolution(x.Tree, x.Qualities), Qualities = x.Qualities }).ToList();
|
---|
[13241] | 133 | nonDominatedSolutions.ForEach(s => s.Solution.Name = string.Join(",", s.Qualities.Select(q => q.ToString())));
|
---|
| 134 |
|
---|
[5557] | 135 | #region update Pareto-optimal solution archive
|
---|
[13241] | 136 | if (nonDominatedSolutions.Count > 0) {
|
---|
| 137 | //add old non-dominated solutions only if they are not dominated by one of the new solutions
|
---|
[5557] | 138 | for (int i = 0; i < trainingBestQualities.Count; i++) {
|
---|
[13241] | 139 | if (IsNonDominated(trainingBestQualities[i], nonDominatedSolutions.Select(x => x.Qualities), maximization)) {
|
---|
| 140 | nonDominatedSolutions.Add(new { Solution = TrainingBestSolutions[i], Qualities = TrainingBestSolutionQualities[i].ToArray() });
|
---|
[5557] | 141 | }
|
---|
| 142 | }
|
---|
| 143 |
|
---|
[13241] | 144 | //assumes the the first objective is always the accuracy
|
---|
| 145 | var sortedNonDominatedSolutions = maximization[0]
|
---|
| 146 | ? nonDominatedSolutions.OrderByDescending(x => x.Qualities[0])
|
---|
| 147 | : nonDominatedSolutions.OrderBy(x => x.Qualities[0]);
|
---|
| 148 | var trainingBestSolution = sortedNonDominatedSolutions.Select(s => s.Solution).First();
|
---|
| 149 | results[TrainingBestSolutionParameterName].Value = trainingBestSolution;
|
---|
| 150 | TrainingBestSolutions = new ItemList<T>(sortedNonDominatedSolutions.Select(x => x.Solution));
|
---|
| 151 | results[TrainingBestSolutionsParameter.Name].Value = TrainingBestSolutions;
|
---|
| 152 | TrainingBestSolutionQualities = new ItemList<DoubleArray>(sortedNonDominatedSolutions.Select(x => new DoubleArray(x.Qualities)));
|
---|
| 153 | results[TrainingBestSolutionQualitiesParameter.Name].Value = TrainingBestSolutionQualities;
|
---|
[5557] | 154 | }
|
---|
| 155 | #endregion
|
---|
| 156 | return base.Apply();
|
---|
| 157 | }
|
---|
| 158 |
|
---|
| 159 | protected abstract T CreateSolution(ISymbolicExpressionTree bestTree, double[] bestQuality);
|
---|
| 160 |
|
---|
[13241] | 161 | private bool IsNonDominated(double[] point, IEnumerable<double[]> points, bool[] maximization) {
|
---|
[5557] | 162 | foreach (var refPoint in points) {
|
---|
[13241] | 163 | bool refPointDominatesPoint = IsBetterOrEqual(refPoint, point, maximization);
|
---|
[5557] | 164 | if (refPointDominatesPoint) return false;
|
---|
| 165 | }
|
---|
| 166 | return true;
|
---|
| 167 | }
|
---|
[13241] | 168 |
|
---|
| 169 | private bool IsBetterOrEqual(double[] lhs, double[] rhs, bool[] maximization) {
|
---|
| 170 | for (int i = 0; i < lhs.Length; i++) {
|
---|
| 171 | var result = IsBetterOrEqual(lhs[i], rhs[i], maximization[i]);
|
---|
| 172 | if (!result) return false;
|
---|
| 173 | }
|
---|
| 174 | return true;
|
---|
| 175 | }
|
---|
| 176 |
|
---|
[5742] | 177 | private bool IsBetterOrEqual(double lhs, double rhs, bool maximization) {
|
---|
[13241] | 178 | if (maximization) return lhs >= rhs;
|
---|
| 179 | else return lhs <= rhs;
|
---|
[5557] | 180 | }
|
---|
| 181 | }
|
---|
| 182 | }
|
---|