1 | using System;
|
---|
2 | using System.Collections.Generic;
|
---|
3 | using System.Linq;
|
---|
4 | using System.Threading;
|
---|
5 | using HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration.GrammarEnumeration;
|
---|
6 | using HeuristicLab.Collections;
|
---|
7 | using HeuristicLab.Common;
|
---|
8 | using HeuristicLab.Core;
|
---|
9 | using HeuristicLab.Data;
|
---|
10 | using HeuristicLab.Optimization;
|
---|
11 | using HeuristicLab.Parameters;
|
---|
12 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
13 | using HeuristicLab.Problems.DataAnalysis;
|
---|
14 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
15 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;
|
---|
16 |
|
---|
17 | namespace HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration {
|
---|
18 | [Item("Grammar Enumeration Symbolic Regression", "Iterates all possible model structures for a fixed grammar.")]
|
---|
19 | [StorableClass]
|
---|
20 | [Creatable(CreatableAttribute.Categories.DataAnalysisRegression, Priority = 250)]
|
---|
21 | public class GrammarEnumerationAlgorithm : FixedDataAnalysisAlgorithm<IRegressionProblem> {
|
---|
22 | #region properties and result names
|
---|
23 | private readonly string VariableImportanceWeightName = "Variable Importance Weight";
|
---|
24 | private readonly string SearchStructureSizeName = "Search Structure Size";
|
---|
25 | private readonly string GeneratedPhrasesName = "Generated/Archived Phrases";
|
---|
26 | private readonly string GeneratedSentencesName = "Generated Sentences";
|
---|
27 | private readonly string DistinctSentencesName = "Distinct Sentences";
|
---|
28 | private readonly string PhraseExpansionsName = "Phrase Expansions";
|
---|
29 | private readonly string AverageSentenceComplexityName = "Avg. Sentence Complexity among Distinct";
|
---|
30 | private readonly string OverwrittenSentencesName = "Sentences overwritten";
|
---|
31 | private readonly string AnalyzersParameterName = "Analyzers";
|
---|
32 | private readonly string ExpansionsPerSecondName = "Expansions per second";
|
---|
33 |
|
---|
34 | private readonly string OptimizeConstantsParameterName = "Optimize Constants";
|
---|
35 | private readonly string ErrorWeightParameterName = "Error Weight";
|
---|
36 | private readonly string SearchDataStructureParameterName = "Search Data Structure";
|
---|
37 | private readonly string MaxComplexityParameterName = "Max. Complexity";
|
---|
38 | private readonly string GuiUpdateIntervalParameterName = "GUI Update Interval";
|
---|
39 | private readonly string GrammarSymbolsParameterName = "Grammar Symbols";
|
---|
40 | private readonly string SearchCacheSizeParameterName = "Search Cache Size";
|
---|
41 | private readonly string SearchDataStructureSizeParameterName = "Search Data Structure Size";
|
---|
42 |
|
---|
43 | // result names
|
---|
44 | public static readonly string BestTrainingModelResultName = "Best model (Training)";
|
---|
45 | public static readonly string BestTrainingSolutionResultName = "Best solution (Training)";
|
---|
46 | public static readonly string BestComplexityResultName = "Best solution complexity";
|
---|
47 |
|
---|
48 | public override bool SupportsPause { get { return true; } }
|
---|
49 |
|
---|
50 | protected IFixedValueParameter<DoubleValue> VariableImportanceWeightParameter {
|
---|
51 | get { return (IFixedValueParameter<DoubleValue>)Parameters[VariableImportanceWeightName]; }
|
---|
52 | }
|
---|
53 |
|
---|
54 | public double VariableImportanceWeight {
|
---|
55 | get { return VariableImportanceWeightParameter.Value.Value; }
|
---|
56 | set { VariableImportanceWeightParameter.Value.Value = value; }
|
---|
57 | }
|
---|
58 |
|
---|
59 | protected IFixedValueParameter<BoolValue> OptimizeConstantsParameter {
|
---|
60 | get { return (IFixedValueParameter<BoolValue>)Parameters[OptimizeConstantsParameterName]; }
|
---|
61 | }
|
---|
62 |
|
---|
63 | public bool OptimizeConstants {
|
---|
64 | get { return OptimizeConstantsParameter.Value.Value; }
|
---|
65 | set { OptimizeConstantsParameter.Value.Value = value; }
|
---|
66 | }
|
---|
67 |
|
---|
68 | protected IFixedValueParameter<IntValue> MaxComplexityParameter {
|
---|
69 | get { return (IFixedValueParameter<IntValue>)Parameters[MaxComplexityParameterName]; }
|
---|
70 | }
|
---|
71 |
|
---|
72 | public int MaxComplexity {
|
---|
73 | get { return MaxComplexityParameter.Value.Value; }
|
---|
74 | set { MaxComplexityParameter.Value.Value = value; }
|
---|
75 | }
|
---|
76 |
|
---|
77 | protected IFixedValueParameter<DoubleValue> ErrorWeightParameter {
|
---|
78 | get { return (IFixedValueParameter<DoubleValue>)Parameters[ErrorWeightParameterName]; }
|
---|
79 | }
|
---|
80 |
|
---|
81 | public double ErrorWeight {
|
---|
82 | get { return ErrorWeightParameter.Value.Value; }
|
---|
83 | set { ErrorWeightParameter.Value.Value = value; }
|
---|
84 | }
|
---|
85 |
|
---|
86 | protected IFixedValueParameter<IntValue> GuiUpdateIntervalParameter {
|
---|
87 | get { return (IFixedValueParameter<IntValue>)Parameters[GuiUpdateIntervalParameterName]; }
|
---|
88 | }
|
---|
89 |
|
---|
90 | public int GuiUpdateInterval {
|
---|
91 | get { return GuiUpdateIntervalParameter.Value.Value; }
|
---|
92 | set { GuiUpdateIntervalParameter.Value.Value = value; }
|
---|
93 | }
|
---|
94 |
|
---|
95 | protected IFixedValueParameter<EnumValue<StorageType>> SearchDataStructureParameter {
|
---|
96 | get { return (IFixedValueParameter<EnumValue<StorageType>>)Parameters[SearchDataStructureParameterName]; }
|
---|
97 | }
|
---|
98 |
|
---|
99 | public IFixedValueParameter<IntValue> SearchDataStructureSizeParameter {
|
---|
100 | get { return (IFixedValueParameter<IntValue>)Parameters[SearchDataStructureSizeParameterName]; }
|
---|
101 | }
|
---|
102 |
|
---|
103 | public int SearchDataStructureSize {
|
---|
104 | get { return SearchDataStructureSizeParameter.Value.Value; }
|
---|
105 | }
|
---|
106 |
|
---|
107 | public IFixedValueParameter<IntValue> SearchCacheSizeParameter {
|
---|
108 | get { return (IFixedValueParameter<IntValue>)Parameters[SearchCacheSizeParameterName]; }
|
---|
109 | }
|
---|
110 |
|
---|
111 | public int SearchCacheSize {
|
---|
112 | get { return SearchCacheSizeParameter.Value.Value; }
|
---|
113 | }
|
---|
114 |
|
---|
115 | public StorageType SearchDataStructure {
|
---|
116 | get { return SearchDataStructureParameter.Value.Value; }
|
---|
117 | set { SearchDataStructureParameter.Value.Value = value; }
|
---|
118 | }
|
---|
119 |
|
---|
120 | public IFixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>> AnalyzersParameter {
|
---|
121 | get { return (IFixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>>)Parameters[AnalyzersParameterName]; }
|
---|
122 | }
|
---|
123 |
|
---|
124 | public ICheckedItemCollection<IGrammarEnumerationAnalyzer> Analyzers {
|
---|
125 | get { return AnalyzersParameter.Value; }
|
---|
126 | }
|
---|
127 |
|
---|
128 | public IFixedValueParameter<ReadOnlyCheckedItemCollection<EnumValue<GrammarRule>>> GrammarSymbolsParameter {
|
---|
129 | get { return (IFixedValueParameter<ReadOnlyCheckedItemCollection<EnumValue<GrammarRule>>>)Parameters[GrammarSymbolsParameterName]; }
|
---|
130 | }
|
---|
131 |
|
---|
132 | public ReadOnlyCheckedItemCollection<EnumValue<GrammarRule>> GrammarSymbols {
|
---|
133 | get { return GrammarSymbolsParameter.Value; }
|
---|
134 | }
|
---|
135 |
|
---|
136 | [Storable]
|
---|
137 | public SymbolString BestTrainingSentence { get; set; } // Currently set in RSquaredEvaluator: quite hacky, but makes testing much easier for now...
|
---|
138 | #endregion
|
---|
139 |
|
---|
140 | [Storable]
|
---|
141 | public Dictionary<int, int> DistinctSentencesComplexity { get; private set; } // Semantically distinct sentences and their length in a run.
|
---|
142 |
|
---|
143 | [Storable]
|
---|
144 | public HashSet<int> ArchivedPhrases { get; private set; }
|
---|
145 |
|
---|
146 | [Storable]
|
---|
147 | internal SearchDataStore OpenPhrases { get; private set; } // Stack/Queue/etc. for fetching the next node in the search tree.
|
---|
148 |
|
---|
149 | #region execution stats
|
---|
150 | [Storable]
|
---|
151 | public int AllGeneratedSentencesCount { get; private set; }
|
---|
152 |
|
---|
153 | [Storable]
|
---|
154 | public int OverwrittenSentencesCount { get; private set; } // It is not guaranteed that shorter solutions are found first.
|
---|
155 | // When longer solutions are overwritten with shorter ones,
|
---|
156 | // this counter is increased.
|
---|
157 | [Storable]
|
---|
158 | public int PhraseExpansionCount { get; private set; } // Number, how many times a nonterminal symbol is replaced with a production rule.
|
---|
159 | #endregion
|
---|
160 |
|
---|
161 | [Storable]
|
---|
162 | public Grammar Grammar { get; private set; }
|
---|
163 |
|
---|
164 | #region ctors
|
---|
165 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
166 | return new GrammarEnumerationAlgorithm(this, cloner);
|
---|
167 | }
|
---|
168 |
|
---|
169 | [StorableConstructor]
|
---|
170 | protected GrammarEnumerationAlgorithm(bool deserializing) : base(deserializing) { }
|
---|
171 |
|
---|
172 | public GrammarEnumerationAlgorithm() {
|
---|
173 | Parameters.Add(new FixedValueParameter<DoubleValue>(VariableImportanceWeightName, "Variable Weight.", new DoubleValue(1.0)));
|
---|
174 | Parameters.Add(new FixedValueParameter<BoolValue>(OptimizeConstantsParameterName, "Run constant optimization in sentence evaluation.", new BoolValue(false)));
|
---|
175 | Parameters.Add(new FixedValueParameter<DoubleValue>(ErrorWeightParameterName, "Defines, how much weight is put on a phrase's r² value when priorizing phrases during search.", new DoubleValue(0.8)));
|
---|
176 | Parameters.Add(new FixedValueParameter<IntValue>(MaxComplexityParameterName, "The maximum number of variable symbols in a sentence.", new IntValue(12)));
|
---|
177 | Parameters.Add(new FixedValueParameter<IntValue>(GuiUpdateIntervalParameterName, "Number of generated sentences, until GUI is refreshed.", new IntValue(5000)));
|
---|
178 | Parameters.Add(new FixedValueParameter<IntValue>(SearchCacheSizeParameterName, "The size of the search node cache.", new IntValue((int)1e5)));
|
---|
179 | Parameters.Add(new FixedValueParameter<IntValue>(SearchDataStructureSizeParameterName, "The size of the search data structure.", new IntValue((int)1e5)));
|
---|
180 | Parameters.Add(new FixedValueParameter<EnumValue<StorageType>>(SearchDataStructureParameterName, new EnumValue<StorageType>(StorageType.PriorityQueue)));
|
---|
181 |
|
---|
182 | SearchDataStructureParameter.Value.ValueChanged += (o, e) => Prepare();
|
---|
183 | SearchDataStructureSizeParameter.Value.ValueChanged += (o, e) => Prepare();
|
---|
184 | SearchCacheSizeParameter.Value.ValueChanged += (o, e) => Prepare();
|
---|
185 |
|
---|
186 | var availableAnalyzers = new IGrammarEnumerationAnalyzer[] {
|
---|
187 | new SearchGraphVisualizer(),
|
---|
188 | new SentenceLogger(),
|
---|
189 | new RSquaredEvaluator()
|
---|
190 | };
|
---|
191 |
|
---|
192 | Parameters.Add(new FixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>>(
|
---|
193 | AnalyzersParameterName,
|
---|
194 | new CheckedItemCollection<IGrammarEnumerationAnalyzer>(availableAnalyzers).AsReadOnly()));
|
---|
195 |
|
---|
196 | Analyzers.CheckedItemsChanged += Analyzers_CheckedItemsChanged;
|
---|
197 |
|
---|
198 | foreach (var analyzer in Analyzers) {
|
---|
199 | Analyzers.SetItemCheckedState(analyzer, false);
|
---|
200 | }
|
---|
201 | Analyzers.SetItemCheckedState(Analyzers.First(analyzer => analyzer is RSquaredEvaluator), true);
|
---|
202 |
|
---|
203 | var grammarSymbols = Enum.GetValues(typeof(GrammarRule))
|
---|
204 | .Cast<GrammarRule>()
|
---|
205 | .Select(v => new EnumValue<GrammarRule>(v));
|
---|
206 |
|
---|
207 | Parameters.Add(new FixedValueParameter<ReadOnlyCheckedItemCollection<EnumValue<GrammarRule>>>(
|
---|
208 | GrammarSymbolsParameterName,
|
---|
209 | new ReadOnlyCheckedItemCollection<EnumValue<GrammarRule>>(new CheckedItemCollection<EnumValue<GrammarRule>>(grammarSymbols))
|
---|
210 | ));
|
---|
211 | foreach (EnumValue<GrammarRule> grammarSymbol in GrammarSymbols) {
|
---|
212 | GrammarSymbols.SetItemCheckedState(grammarSymbol, true);
|
---|
213 | }
|
---|
214 |
|
---|
215 | // set a default problem
|
---|
216 | Problem = new RegressionProblem() {
|
---|
217 | ProblemData = new Problems.Instances.DataAnalysis.PolyTen(seed: 1234).GenerateRegressionData()
|
---|
218 | };
|
---|
219 | }
|
---|
220 |
|
---|
221 | public GrammarEnumerationAlgorithm(GrammarEnumerationAlgorithm original, Cloner cloner) : base(original, cloner) {
|
---|
222 | foreach (var analyzer in Analyzers.CheckedItems)
|
---|
223 | analyzer.Register(this);
|
---|
224 | Analyzers.CheckedItemsChanged += Analyzers_CheckedItemsChanged;
|
---|
225 |
|
---|
226 | DistinctSentencesComplexity = new Dictionary<int, int>(original.DistinctSentencesComplexity);
|
---|
227 | ArchivedPhrases = new HashSet<int>(original.ArchivedPhrases);
|
---|
228 | OpenPhrases = cloner.Clone(original.OpenPhrases);
|
---|
229 | Grammar = cloner.Clone(original.Grammar);
|
---|
230 |
|
---|
231 | AllGeneratedSentencesCount = original.AllGeneratedSentencesCount;
|
---|
232 | OverwrittenSentencesCount = original.OverwrittenSentencesCount;
|
---|
233 | PhraseExpansionCount = original.PhraseExpansionCount;
|
---|
234 |
|
---|
235 | if (original.variableImportance != null)
|
---|
236 | variableImportance = new Dictionary<VariableTerminalSymbol, double>(original.variableImportance);
|
---|
237 | }
|
---|
238 | #endregion
|
---|
239 |
|
---|
240 | [Storable]
|
---|
241 | private Dictionary<VariableTerminalSymbol, double> variableImportance;
|
---|
242 |
|
---|
243 | public override void Prepare() {
|
---|
244 | DistinctSentencesComplexity = new Dictionary<int, int>();
|
---|
245 | ArchivedPhrases = new HashSet<int>();
|
---|
246 | AllGeneratedSentencesCount = 0;
|
---|
247 | OverwrittenSentencesCount = 0;
|
---|
248 | PhraseExpansionCount = 0;
|
---|
249 |
|
---|
250 | Analyzers.OfType<RSquaredEvaluator>().First().OptimizeConstants = OptimizeConstants;
|
---|
251 | Grammar = new Grammar(Problem.ProblemData.AllowedInputVariables.ToArray(), GrammarSymbols.CheckedItems.Select(v => v.Value));
|
---|
252 | OpenPhrases = new SearchDataStore(SearchDataStructure, SearchDataStructureSize, SearchCacheSize); // Select search strategy
|
---|
253 |
|
---|
254 | CalculateVariableImportances();
|
---|
255 |
|
---|
256 | base.Prepare(); // this actually clears the results which will get reinitialized on Run()
|
---|
257 | }
|
---|
258 |
|
---|
259 | private void CalculateVariableImportances() {
|
---|
260 | variableImportance = new Dictionary<VariableTerminalSymbol, double>();
|
---|
261 |
|
---|
262 | RandomForestRegression rf = new RandomForestRegression();
|
---|
263 | rf.Problem = Problem;
|
---|
264 | rf.Start();
|
---|
265 | IRegressionSolution rfSolution = (RandomForestRegressionSolution)rf.Results["Random forest regression solution"].Value;
|
---|
266 | var rfImpacts = RegressionSolutionVariableImpactsCalculator.CalculateImpacts(
|
---|
267 | rfSolution,
|
---|
268 | RegressionSolutionVariableImpactsCalculator.DataPartitionEnum.Training,
|
---|
269 | RegressionSolutionVariableImpactsCalculator.ReplacementMethodEnum.Shuffle);
|
---|
270 |
|
---|
271 | // save the normalized importances
|
---|
272 | var sum = rfImpacts.Sum(x => x.Item2);
|
---|
273 | foreach (Tuple<string, double> rfImpact in rfImpacts) {
|
---|
274 | VariableTerminalSymbol varSym = Grammar.VarTerminals.First(v => v.StringRepresentation == rfImpact.Item1);
|
---|
275 | variableImportance[varSym] = rfImpact.Item2 / sum;
|
---|
276 | }
|
---|
277 | }
|
---|
278 |
|
---|
279 | protected override void Run(CancellationToken cancellationToken) {
|
---|
280 | // do not reinitialize the algorithm if we're resuming from pause
|
---|
281 | if (previousExecutionState != ExecutionState.Paused) {
|
---|
282 | InitResults();
|
---|
283 | var phrase0 = new SymbolString(new[] { Grammar.StartSymbol });
|
---|
284 | var phrase0Hash = Grammar.Hasher.CalcHashCode(phrase0);
|
---|
285 |
|
---|
286 | OpenPhrases.Store(new SearchNode(phrase0Hash, 0.0, 0.0, phrase0));
|
---|
287 | }
|
---|
288 |
|
---|
289 | int maxSentenceLength = GetMaxSentenceLength();
|
---|
290 | var errorWeight = ErrorWeight;
|
---|
291 | var variableImportanceWeight = VariableImportanceWeight;
|
---|
292 | // main search loop
|
---|
293 | while (OpenPhrases.Count > 0) {
|
---|
294 | if (cancellationToken.IsCancellationRequested)
|
---|
295 | break;
|
---|
296 |
|
---|
297 | SearchNode fetchedSearchNode = OpenPhrases.GetNext();
|
---|
298 |
|
---|
299 | if (fetchedSearchNode == null)
|
---|
300 | continue;
|
---|
301 |
|
---|
302 | SymbolString currPhrase = fetchedSearchNode.SymbolString;
|
---|
303 |
|
---|
304 | OnPhraseFetched(fetchedSearchNode.Hash, currPhrase);
|
---|
305 |
|
---|
306 | ArchivedPhrases.Add(fetchedSearchNode.Hash);
|
---|
307 |
|
---|
308 | // expand next nonterminal symbols
|
---|
309 | int nonterminalSymbolIndex = currPhrase.NextNonterminalIndex();
|
---|
310 | NonterminalSymbol expandedSymbol = (NonterminalSymbol)currPhrase[nonterminalSymbolIndex];
|
---|
311 | var appliedProductions = Grammar.Productions[expandedSymbol];
|
---|
312 |
|
---|
313 | for (int i = 0; i < appliedProductions.Count; i++) {
|
---|
314 | PhraseExpansionCount++;
|
---|
315 |
|
---|
316 | SymbolString newPhrase = currPhrase.DerivePhrase(nonterminalSymbolIndex, appliedProductions[i]);
|
---|
317 | int newPhraseComplexity = Grammar.GetComplexity(newPhrase);
|
---|
318 |
|
---|
319 | if (newPhraseComplexity > MaxComplexity)
|
---|
320 | continue;
|
---|
321 |
|
---|
322 | var phraseHash = Grammar.Hasher.CalcHashCode(newPhrase);
|
---|
323 |
|
---|
324 | OnPhraseDerived(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
|
---|
325 |
|
---|
326 | if (newPhrase.IsSentence()) {
|
---|
327 | AllGeneratedSentencesCount++;
|
---|
328 |
|
---|
329 | OnSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
|
---|
330 |
|
---|
331 | // Is the best solution found? (only if RSquaredEvaluator is activated)
|
---|
332 | if (Results.ContainsKey(RSquaredEvaluator.BestTrainingQualityResultName)) {
|
---|
333 | double r2 = ((DoubleValue)Results[RSquaredEvaluator.BestTrainingQualityResultName].Value).Value;
|
---|
334 | if (r2.IsAlmost(1.0)) {
|
---|
335 | UpdateView(force: true);
|
---|
336 | return;
|
---|
337 | }
|
---|
338 | }
|
---|
339 |
|
---|
340 | if (!DistinctSentencesComplexity.ContainsKey(phraseHash) || DistinctSentencesComplexity[phraseHash] > newPhraseComplexity) {
|
---|
341 | if (DistinctSentencesComplexity.ContainsKey(phraseHash)) OverwrittenSentencesCount++; // for analysis only
|
---|
342 |
|
---|
343 | DistinctSentencesComplexity[phraseHash] = newPhraseComplexity;
|
---|
344 | OnDistinctSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
|
---|
345 | }
|
---|
346 | UpdateView();
|
---|
347 |
|
---|
348 | } else if (!OpenPhrases.Contains(phraseHash) && !ArchivedPhrases.Contains(phraseHash)) {
|
---|
349 |
|
---|
350 | double r2 = GetR2(newPhrase, fetchedSearchNode.R2);
|
---|
351 | double phrasePriority = GetPriority(newPhrase, r2, maxSentenceLength, errorWeight, variableImportanceWeight);
|
---|
352 |
|
---|
353 | SearchNode newSearchNode = new SearchNode(phraseHash, phrasePriority, r2, newPhrase);
|
---|
354 | OpenPhrases.Store(newSearchNode);
|
---|
355 | }
|
---|
356 | }
|
---|
357 | }
|
---|
358 | UpdateView(force: true);
|
---|
359 | }
|
---|
360 |
|
---|
361 | protected double GetPriority(SymbolString phrase, double r2, int maxSentenceLength, double errorWeight, double variableImportanceWeight) {
|
---|
362 | var distinctVars = phrase.OfType<VariableTerminalSymbol>().Distinct();
|
---|
363 |
|
---|
364 | var sum = 0d;
|
---|
365 | foreach (var v in distinctVars) {
|
---|
366 | sum += variableImportance[v];
|
---|
367 | }
|
---|
368 | var phraseVariableImportance = 1 - sum;
|
---|
369 |
|
---|
370 | double relLength = (double)phrase.Count() / maxSentenceLength;
|
---|
371 | double error = 1.0 - r2;
|
---|
372 |
|
---|
373 | return relLength + errorWeight * error + variableImportanceWeight * phraseVariableImportance;
|
---|
374 | }
|
---|
375 |
|
---|
376 | private double GetR2(SymbolString phrase, double parentR2) {
|
---|
377 | int length = phrase.Count();
|
---|
378 |
|
---|
379 | // If the only nonterminal symbol is Expr, we can need to evaluate the sentence. Otherwise
|
---|
380 | // the phrase has the same r2 as its parent, from which it was derived.
|
---|
381 | for (int i = 0; i < length; i++) {
|
---|
382 | if (phrase[i] is NonterminalSymbol && phrase[i] != Grammar.Expr) {
|
---|
383 | return parentR2;
|
---|
384 | }
|
---|
385 | }
|
---|
386 |
|
---|
387 | return Grammar.EvaluatePhrase(phrase, Problem.ProblemData, OptimizeConstants);
|
---|
388 | }
|
---|
389 |
|
---|
390 | private int GetMaxSentenceLength() {
|
---|
391 | SymbolString s = new SymbolString(Grammar.StartSymbol);
|
---|
392 |
|
---|
393 | while (!s.IsSentence() && Grammar.GetComplexity(s) <= MaxComplexity) {
|
---|
394 | int expandedSymbolIndex = s.NextNonterminalIndex();
|
---|
395 | NonterminalSymbol expandedSymbol = (NonterminalSymbol)s[expandedSymbolIndex];
|
---|
396 |
|
---|
397 | var productions = Grammar.Productions[expandedSymbol];
|
---|
398 | var longestProduction = productions // Find production with most terminal symbols to expand as much as possible...
|
---|
399 | .OrderBy(CountTerminals) // but with lowest complexity/nonterminal count to keep complexity low.
|
---|
400 | .ThenByDescending(CountNonTerminals)
|
---|
401 | .First();
|
---|
402 |
|
---|
403 | s = s.DerivePhrase(expandedSymbolIndex, longestProduction);
|
---|
404 | }
|
---|
405 |
|
---|
406 | return s.Count();
|
---|
407 | }
|
---|
408 |
|
---|
409 | private int CountTerminals(Production p) {
|
---|
410 | return p.Count(s => s is TerminalSymbol);
|
---|
411 | }
|
---|
412 |
|
---|
413 | private int CountNonTerminals(Production p) {
|
---|
414 | return p.Count(s => s is NonterminalSymbol);
|
---|
415 | }
|
---|
416 |
|
---|
417 | #region pause support
|
---|
418 | private ExecutionState previousExecutionState;
|
---|
419 | protected override void OnPaused() {
|
---|
420 | previousExecutionState = this.ExecutionState;
|
---|
421 | base.OnPaused();
|
---|
422 | }
|
---|
423 | protected override void OnPrepared() {
|
---|
424 | previousExecutionState = this.ExecutionState;
|
---|
425 | base.OnPrepared();
|
---|
426 | }
|
---|
427 | protected override void OnStarted() {
|
---|
428 | previousExecutionState = this.ExecutionState;
|
---|
429 | base.OnStarted();
|
---|
430 | }
|
---|
431 | protected override void OnStopped() {
|
---|
432 | previousExecutionState = this.ExecutionState;
|
---|
433 |
|
---|
434 | if (BestTrainingSentence == null) {
|
---|
435 | base.OnStopped();
|
---|
436 | return;
|
---|
437 | }
|
---|
438 |
|
---|
439 | var tree = Grammar.ParseSymbolicExpressionTree(BestTrainingSentence);
|
---|
440 | var model = new SymbolicRegressionModel(Problem.ProblemData.TargetVariable, tree, new SymbolicDataAnalysisExpressionTreeLinearInterpreter());
|
---|
441 | model.Scale(Problem.ProblemData);
|
---|
442 | var bestTrainingSolution = new SymbolicRegressionSolution(model, Problem.ProblemData);
|
---|
443 | Results.AddOrUpdateResult(BestTrainingModelResultName, model);
|
---|
444 | Results.AddOrUpdateResult(BestTrainingSolutionResultName, bestTrainingSolution);
|
---|
445 | Results.AddOrUpdateResult(BestComplexityResultName, new IntValue(Grammar.GetComplexity(BestTrainingSentence)));
|
---|
446 | base.OnStopped();
|
---|
447 | }
|
---|
448 | #endregion
|
---|
449 |
|
---|
450 | #region Visualization in HL
|
---|
451 | // Initialize entries in result set.
|
---|
452 | private void InitResults() {
|
---|
453 | Results.Clear();
|
---|
454 | Results.Add(new Result(GeneratedPhrasesName, new IntValue(0)));
|
---|
455 | Results.Add(new Result(SearchStructureSizeName, new IntValue(0)));
|
---|
456 | Results.Add(new Result(GeneratedSentencesName, new IntValue(0)));
|
---|
457 | Results.Add(new Result(DistinctSentencesName, new IntValue(0)));
|
---|
458 | Results.Add(new Result(PhraseExpansionsName, new IntValue(0)));
|
---|
459 | Results.Add(new Result(OverwrittenSentencesName, new IntValue(0)));
|
---|
460 | Results.Add(new Result(AverageSentenceComplexityName, new DoubleValue(1.0)));
|
---|
461 | Results.Add(new Result(ExpansionsPerSecondName, "In Thousand expansions per second", new IntValue(0)));
|
---|
462 | }
|
---|
463 |
|
---|
464 | // Update the view for intermediate results in an algorithm run.
|
---|
465 | private int updates;
|
---|
466 | private void UpdateView(bool force = false) {
|
---|
467 | updates++;
|
---|
468 |
|
---|
469 | if (force || updates % GuiUpdateInterval == 1) {
|
---|
470 | ((IntValue)Results[GeneratedPhrasesName].Value).Value = ArchivedPhrases.Count;
|
---|
471 | ((IntValue)Results[SearchStructureSizeName].Value).Value = OpenPhrases.Count;
|
---|
472 | ((IntValue)Results[GeneratedSentencesName].Value).Value = AllGeneratedSentencesCount;
|
---|
473 | ((IntValue)Results[DistinctSentencesName].Value).Value = DistinctSentencesComplexity.Count;
|
---|
474 | ((IntValue)Results[PhraseExpansionsName].Value).Value = PhraseExpansionCount;
|
---|
475 | ((DoubleValue)Results[AverageSentenceComplexityName].Value).Value = DistinctSentencesComplexity.Count > 0
|
---|
476 | ? DistinctSentencesComplexity.Select(pair => pair.Value).Average()
|
---|
477 | : 0;
|
---|
478 | ((IntValue)Results[OverwrittenSentencesName].Value).Value = OverwrittenSentencesCount;
|
---|
479 | ((IntValue)Results[ExpansionsPerSecondName].Value).Value = (int)((PhraseExpansionCount /
|
---|
480 | ExecutionTime.TotalSeconds) / 1000.0);
|
---|
481 | }
|
---|
482 | }
|
---|
483 | #endregion
|
---|
484 |
|
---|
485 | #region events
|
---|
486 |
|
---|
487 | // private event handlers for analyzers
|
---|
488 | private void Analyzers_CheckedItemsChanged(object sender, CollectionItemsChangedEventArgs<IGrammarEnumerationAnalyzer> args) {
|
---|
489 | // newly added items
|
---|
490 | foreach (var item in args.Items.Except(args.OldItems).Union(args.OldItems.Except(args.Items))) {
|
---|
491 | if (Analyzers.ItemChecked(item)) {
|
---|
492 | item.Register(this);
|
---|
493 | } else {
|
---|
494 | item.Deregister(this);
|
---|
495 | }
|
---|
496 | }
|
---|
497 | }
|
---|
498 |
|
---|
499 | public event EventHandler<PhraseEventArgs> PhraseFetched;
|
---|
500 | private void OnPhraseFetched(int hash, SymbolString symbolString) {
|
---|
501 | if (PhraseFetched != null) {
|
---|
502 | PhraseFetched(this, new PhraseEventArgs(hash, symbolString));
|
---|
503 | }
|
---|
504 | }
|
---|
505 |
|
---|
506 | public event EventHandler<PhraseAddedEventArgs> PhraseDerived;
|
---|
507 | private void OnPhraseDerived(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
|
---|
508 | if (PhraseDerived != null) {
|
---|
509 | PhraseDerived(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
|
---|
510 | }
|
---|
511 | }
|
---|
512 |
|
---|
513 | public event EventHandler<PhraseAddedEventArgs> SentenceGenerated;
|
---|
514 | private void OnSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
|
---|
515 | if (SentenceGenerated != null) {
|
---|
516 | SentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
|
---|
517 | }
|
---|
518 | }
|
---|
519 |
|
---|
520 | public event EventHandler<PhraseAddedEventArgs> DistinctSentenceGenerated;
|
---|
521 | private void OnDistinctSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
|
---|
522 | if (DistinctSentenceGenerated != null) {
|
---|
523 | DistinctSentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
|
---|
524 | }
|
---|
525 | }
|
---|
526 |
|
---|
527 | #endregion
|
---|
528 |
|
---|
529 | }
|
---|
530 |
|
---|
531 | #region events for analysis
|
---|
532 |
|
---|
533 | public class PhraseEventArgs : EventArgs {
|
---|
534 | public int Hash { get; }
|
---|
535 |
|
---|
536 | public SymbolString Phrase { get; }
|
---|
537 |
|
---|
538 | public PhraseEventArgs(int hash, SymbolString phrase) {
|
---|
539 | Hash = hash;
|
---|
540 | Phrase = phrase;
|
---|
541 | }
|
---|
542 | }
|
---|
543 |
|
---|
544 | public class PhraseAddedEventArgs : EventArgs {
|
---|
545 | public int ParentHash { get; }
|
---|
546 | public int NewHash { get; }
|
---|
547 |
|
---|
548 | public SymbolString ParentPhrase { get; }
|
---|
549 | public SymbolString NewPhrase { get; }
|
---|
550 |
|
---|
551 | public Symbol ExpandedSymbol { get; }
|
---|
552 |
|
---|
553 | public Production ExpandedProduction { get; }
|
---|
554 |
|
---|
555 | public PhraseAddedEventArgs(int parentHash, SymbolString parentPhrase, int newHash, SymbolString newPhrase, Symbol expandedSymbol, Production expandedProduction) {
|
---|
556 | ParentHash = parentHash;
|
---|
557 | ParentPhrase = parentPhrase;
|
---|
558 | NewHash = newHash;
|
---|
559 | NewPhrase = newPhrase;
|
---|
560 | ExpandedSymbol = expandedSymbol;
|
---|
561 | ExpandedProduction = expandedProduction;
|
---|
562 | }
|
---|
563 | }
|
---|
564 |
|
---|
565 | #endregion
|
---|
566 | } |
---|