1 | using System.Linq;
|
---|
2 | using HeuristicLab.Analysis;
|
---|
3 | using HeuristicLab.Common;
|
---|
4 | using HeuristicLab.Core;
|
---|
5 | using HeuristicLab.Data;
|
---|
6 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
7 | using HeuristicLab.Operators;
|
---|
8 | using HeuristicLab.Optimization;
|
---|
9 | using HeuristicLab.Parameters;
|
---|
10 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
11 | using HeuristicLab.Problems.DataAnalysis;
|
---|
12 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
13 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;
|
---|
14 |
|
---|
15 | namespace HeuristicLab.EvolutionaryTracking {
|
---|
16 | [Item("SymbolicExpressionTreePruningAnalyzer", "An analyzer that identifies introns in the population and optionally removes them.")]
|
---|
17 | [StorableClass]
|
---|
18 | public sealed class SymbolicExpressionTreePruningAnalyzer : SingleSuccessorOperator, IAnalyzer {
|
---|
19 | private const string SymbolicExpressionTreeParameterName = "SymbolicExpressionTree";
|
---|
20 | private const string SymbolicExpressionTreeQualityParameterName = "Quality";
|
---|
21 | private const string UpdateIntervalParameterName = "UpdateInterval";
|
---|
22 | private const string UpdateCounterParameterName = "UpdateCounter";
|
---|
23 | private const string MatchVariablesParameterName = "MatchVariableNames";
|
---|
24 | private const string MatchVariableWeightsParameterName = "MatchVariableWeights";
|
---|
25 | private const string MatchConstantValuesParameterName = "MatchConstantValues";
|
---|
26 | private const string ResultsParameterName = "Results";
|
---|
27 | private const string SymbolicExpressionInterpreterParameterName = "SymbolicExpressionTreeInterpreter";
|
---|
28 | private const string SymbolicRegressionProblemDataParameterName = "ProblemData";
|
---|
29 | private const string FrequencyThresholdParameterName = "FrequencyThreshold";
|
---|
30 | private const string IgnoreSubtreeFrequenciesParameterName = "IgnoreFrequencies";
|
---|
31 |
|
---|
32 | [Storable]
|
---|
33 | private SymbolicRegressionSolutionImpactValuesCalculator calculator;
|
---|
34 | [Storable]
|
---|
35 | private SymbolicExpressionTreeNodeSimilarityComparer comparer;
|
---|
36 |
|
---|
37 | #region Parameters
|
---|
38 | public ValueParameter<IntValue> UpdateIntervalParameter {
|
---|
39 | get { return (ValueParameter<IntValue>)Parameters[UpdateIntervalParameterName]; }
|
---|
40 | }
|
---|
41 | public ValueParameter<IntValue> UpdateCounterParameter {
|
---|
42 | get { return (ValueParameter<IntValue>)Parameters[UpdateCounterParameterName]; }
|
---|
43 | }
|
---|
44 | public ValueParameter<BoolValue> MatchVariableNamesParameter {
|
---|
45 | get { return (ValueParameter<BoolValue>)Parameters[MatchVariablesParameterName]; }
|
---|
46 | }
|
---|
47 | public ValueParameter<BoolValue> MatchVariableWeightsParameter {
|
---|
48 | get { return (ValueParameter<BoolValue>)Parameters[MatchVariableWeightsParameterName]; }
|
---|
49 | }
|
---|
50 | public ValueParameter<BoolValue> MatchConstantValuesParameter {
|
---|
51 | get { return (ValueParameter<BoolValue>)Parameters[MatchConstantValuesParameterName]; }
|
---|
52 | }
|
---|
53 | public IScopeTreeLookupParameter<ISymbolicExpressionTree> SymbolicExpressionTreeParameter {
|
---|
54 | get { return (IScopeTreeLookupParameter<ISymbolicExpressionTree>)Parameters[SymbolicExpressionTreeParameterName]; }
|
---|
55 | }
|
---|
56 | public IScopeTreeLookupParameter<DoubleValue> SymbolicExpressionTreeQualityParameter {
|
---|
57 | get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters[SymbolicExpressionTreeQualityParameterName]; }
|
---|
58 | }
|
---|
59 | public LookupParameter<ResultCollection> ResultsParameter {
|
---|
60 | get { return (LookupParameter<ResultCollection>)Parameters[ResultsParameterName]; }
|
---|
61 | }
|
---|
62 | public ILookupParameter<SymbolicDataAnalysisExpressionTreeInterpreter> SymbolicExpressionTreeInterpreterParameter {
|
---|
63 | get { return (ILookupParameter<SymbolicDataAnalysisExpressionTreeInterpreter>)Parameters[SymbolicExpressionInterpreterParameterName]; }
|
---|
64 | }
|
---|
65 | public ILookupParameter<RegressionProblemData> SymbolicRegressionProblemDataParameter {
|
---|
66 | get { return (ILookupParameter<RegressionProblemData>)Parameters[SymbolicRegressionProblemDataParameterName]; }
|
---|
67 | }
|
---|
68 | public ValueParameter<DoubleValue> FrequencyThresholdParameter {
|
---|
69 | get { return (ValueParameter<DoubleValue>)Parameters[FrequencyThresholdParameterName]; }
|
---|
70 | }
|
---|
71 | public ValueParameter<BoolValue> IgnoreSubtreeFrequenciesParameter {
|
---|
72 | get { return (ValueParameter<BoolValue>)Parameters[IgnoreSubtreeFrequenciesParameterName]; }
|
---|
73 | }
|
---|
74 | #endregion
|
---|
75 |
|
---|
76 | #region Parameter properties
|
---|
77 | public bool MatchValues {
|
---|
78 | get { return MatchConstantValuesParameter.Value.Value; }
|
---|
79 | }
|
---|
80 | public bool MatchWeights {
|
---|
81 | get { return MatchVariableWeightsParameter.Value.Value; }
|
---|
82 | }
|
---|
83 | public bool MatchVariables {
|
---|
84 | get { return MatchVariableNamesParameter.Value.Value; }
|
---|
85 | }
|
---|
86 | public IntValue UpdateCounter {
|
---|
87 | get { return UpdateCounterParameter.Value; }
|
---|
88 | }
|
---|
89 | public IntValue UpdateInterval {
|
---|
90 | get { return UpdateIntervalParameter.Value; }
|
---|
91 | }
|
---|
92 | public ResultCollection Results {
|
---|
93 | get { return ResultsParameter.ActualValue; }
|
---|
94 | }
|
---|
95 | public SymbolicDataAnalysisExpressionTreeInterpreter SymbolicExpressionTreeInterpreter {
|
---|
96 | get { return SymbolicExpressionTreeInterpreterParameter.ActualValue; }
|
---|
97 | }
|
---|
98 | public RegressionProblemData ProblemData {
|
---|
99 | get { return SymbolicRegressionProblemDataParameter.ActualValue; }
|
---|
100 | }
|
---|
101 | public DoubleValue FrequencyThreshold {
|
---|
102 | get { return FrequencyThresholdParameter.Value; }
|
---|
103 | }
|
---|
104 | public bool IgnoreSubtreesFrequencies {
|
---|
105 | get { return IgnoreSubtreeFrequenciesParameter.Value.Value; }
|
---|
106 | }
|
---|
107 | #endregion
|
---|
108 |
|
---|
109 | [StorableConstructor]
|
---|
110 | private SymbolicExpressionTreePruningAnalyzer(bool deserializing) : base(deserializing) { }
|
---|
111 | private SymbolicExpressionTreePruningAnalyzer(SymbolicExpressionTreePruningAnalyzer original, Cloner cloner)
|
---|
112 | : base(original, cloner) {
|
---|
113 | calculator = original.calculator;
|
---|
114 | comparer = original.comparer;
|
---|
115 | }
|
---|
116 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
117 | return new SymbolicExpressionTreePruningAnalyzer(this, cloner);
|
---|
118 | }
|
---|
119 |
|
---|
120 | public SymbolicExpressionTreePruningAnalyzer() {
|
---|
121 | Parameters.Add(new ScopeTreeLookupParameter<ISymbolicExpressionTree>(SymbolicExpressionTreeParameterName, "The symbolic expression trees to analyze."));
|
---|
122 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>(SymbolicExpressionTreeQualityParameterName, "The qualities of the symbolic expression trees"));
|
---|
123 | Parameters.Add(new ValueParameter<IntValue>(UpdateIntervalParameterName, "The interval in which the tree length analysis should be applied.", new IntValue(1)));
|
---|
124 | Parameters.Add(new ValueParameter<IntValue>(UpdateCounterParameterName, "The value which counts how many times the operator was called since the last update", new IntValue(0)));
|
---|
125 | Parameters.Add(new ValueParameter<BoolValue>(MatchVariablesParameterName, "Specify if the symbolic expression tree comparer should match variable names.", new BoolValue(true)));
|
---|
126 | Parameters.Add(new ValueParameter<BoolValue>(MatchVariableWeightsParameterName, "Specify if the symbolic expression tree comparer should match variable weights.", new BoolValue(true)));
|
---|
127 | Parameters.Add(new ValueParameter<BoolValue>(MatchConstantValuesParameterName, "Specify if the symbolic expression tree comparer should match constant values.", new BoolValue(true)));
|
---|
128 | Parameters.Add(new LookupParameter<SymbolicDataAnalysisExpressionTreeInterpreter>(SymbolicExpressionInterpreterParameterName, "Interpreter for symbolic expression trees"));
|
---|
129 | Parameters.Add(new LookupParameter<RegressionProblemData>(SymbolicRegressionProblemDataParameterName, "The symbolic data analysis problem."));
|
---|
130 | Parameters.Add(new ValueLookupParameter<ResultCollection>(ResultsParameterName, "The results collection where the analysis values should be stored."));
|
---|
131 | Parameters.Add(new ValueParameter<DoubleValue>(FrequencyThresholdParameterName, "The frequency threshold a fragment has to exceed before being removed."));
|
---|
132 | Parameters.Add(new ValueParameter<BoolValue>(IgnoreSubtreeFrequenciesParameterName, "Specifies whether or not subtree frequencies should be taken into account."));
|
---|
133 |
|
---|
134 | calculator = new SymbolicRegressionSolutionImpactValuesCalculator();
|
---|
135 | comparer = new SymbolicExpressionTreeNodeSimilarityComparer { MatchConstantValues = MatchValues, MatchVariableWeights = MatchWeights, MatchVariableNames = MatchVariables };
|
---|
136 |
|
---|
137 | UpdateCounterParameter.Hidden = true;
|
---|
138 | UpdateIntervalParameter.Hidden = true;
|
---|
139 | }
|
---|
140 |
|
---|
141 | public override IOperation Apply() {
|
---|
142 | UpdateCounter.Value++;
|
---|
143 | if (UpdateCounter.Value != UpdateInterval.Value) return base.Apply();
|
---|
144 | UpdateCounter.Value = 0;
|
---|
145 |
|
---|
146 | //var prunedSubtrees = new IntValue(0);
|
---|
147 | int prunedNodes = 0;
|
---|
148 | int prunedTrees = 0;
|
---|
149 | DataTable prunedSubtreesTable;
|
---|
150 | if (Results.ContainsKey("Pruning")) {
|
---|
151 | prunedSubtreesTable = (DataTable)Results["Pruning"].Value;
|
---|
152 | } else {
|
---|
153 | prunedSubtreesTable = new DataTable("Pruning");
|
---|
154 | var prunedNodesRow = new DataRow("Pruned Nodes") { VisualProperties = { StartIndexZero = true } };
|
---|
155 | var prunedTreesRow = new DataRow("Pruned Trees") { VisualProperties = { StartIndexZero = true } };
|
---|
156 | prunedSubtreesTable.Rows.Add(prunedNodesRow);
|
---|
157 | prunedSubtreesTable.Rows.Add(prunedTreesRow);
|
---|
158 | Results.Add(new Result("Pruning", prunedSubtreesTable));
|
---|
159 | }
|
---|
160 |
|
---|
161 | var trees = SymbolicExpressionTreeParameter.ActualValue.ToList();
|
---|
162 | var qualities = SymbolicExpressionTreeQualityParameter.ActualValue.ToList();
|
---|
163 | var prunedTreesMap = new bool[trees.Count];
|
---|
164 |
|
---|
165 | for (int i = 0; i != trees.Count; ++i) {
|
---|
166 | var model = new SymbolicRegressionModel(trees[i], SymbolicExpressionTreeInterpreter);
|
---|
167 | var root = trees[i].Root.GetSubtree(0).GetSubtree(0); // skip the program root symbol and the start symbol
|
---|
168 | var nodes = root.IterateNodesPrefix().Select((node, index) => new { Node = node, Index = index }).ToList();
|
---|
169 | for (int j = 0; j < nodes.Count; ++j) {
|
---|
170 | var n = nodes[j];
|
---|
171 | if (n.Node is ConstantTreeNode) continue;
|
---|
172 | double impact = calculator.CalculateImpactValue(model, n.Node, ProblemData, ProblemData.TrainingIndices, qualities[i].Value);
|
---|
173 |
|
---|
174 | //if (impact.IsAlmost(0.0)) {
|
---|
175 | if (impact < 0.001) {
|
---|
176 | bool prune = true;
|
---|
177 | if (!IgnoreSubtreesFrequencies) {
|
---|
178 | double frequency = (double)trees.Count(t => t.Root.ContainsFragment(new Fragment { Root = n.Node, Index = n.Index }, comparer)) / trees.Count;
|
---|
179 | // remove node if its frequency is above the threshold
|
---|
180 | if (frequency < FrequencyThreshold.Value) { prune = false; }
|
---|
181 | }
|
---|
182 | if (!prune) continue;
|
---|
183 | var parent = n.Node.Parent;
|
---|
184 | if (parent != null && !(parent.Symbol is StartSymbol)) {
|
---|
185 | double replacementValue = calculator.CalculateReplacementValue(model, n.Node, ProblemData, ProblemData.TrainingIndices);
|
---|
186 | var constantNode = new ConstantTreeNode(new Constant()) { Value = replacementValue };
|
---|
187 | var index = parent.IndexOfSubtree(n.Node);
|
---|
188 | parent.RemoveSubtree(index);
|
---|
189 | parent.InsertSubtree(index, constantNode);
|
---|
190 | // skip the removed subtree
|
---|
191 | int length = n.Node.GetLength();
|
---|
192 | j += length;
|
---|
193 |
|
---|
194 | if (!prunedTreesMap[i]) {
|
---|
195 | prunedTrees++;
|
---|
196 | prunedTreesMap[i] = true;
|
---|
197 | }
|
---|
198 | prunedNodes += (length - 1);
|
---|
199 | }
|
---|
200 | }
|
---|
201 | }
|
---|
202 | }
|
---|
203 |
|
---|
204 | prunedSubtreesTable.Rows["Pruned Nodes"].Values.Add(prunedNodes);
|
---|
205 | prunedSubtreesTable.Rows["Pruned Trees"].Values.Add(prunedTrees);
|
---|
206 |
|
---|
207 | return base.Apply();
|
---|
208 | }
|
---|
209 |
|
---|
210 | public bool EnabledByDefault { get { return true; } }
|
---|
211 |
|
---|
212 | }
|
---|
213 | }
|
---|