Changeset 16026 for branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs

Timestamp:

07/27/18 19:28:18 (6 years ago)

Author:

bburlacu

Message:

#2886:

• replace functionally-overlapping classes Production and SymbolString with a single class SymbolList
• refactor methods from Grammar class as methods and properties of SymbolList
• add parameter for the number of constant optimization iterations
• refactor code

File:

• branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs (modified) (18 diffs)

branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs

@@ -32 +32 @@
 
     private readonly string OptimizeConstantsParameterName = "Optimize Constants";
+    private readonly string ConstantOptimizationIterationsParameterName = "Constant Optimization Iterations";
     private readonly string ErrorWeightParameterName = "Error Weight";
     private readonly string SearchDataStructureParameterName = "Search Data Structure";
@@ -55 +56 @@
     }
 
+    public int ConstantOptimizationIterations {
+      get { return ConstantOptimizationIterationsParameter.Value.Value; }
+      set { ConstantOptimizationIterationsParameter.Value.Value = value; }
+    }
+
+    public IFixedValueParameter<IntValue> ConstantOptimizationIterationsParameter {
+      get { return (IFixedValueParameter<IntValue>)Parameters[ConstantOptimizationIterationsParameterName]; }
+    }
+
     protected IFixedValueParameter<IntValue> MaxComplexityParameter {
       get { return (IFixedValueParameter<IntValue>)Parameters[MaxComplexityParameterName]; }
@@ -116 +126 @@
 
     [Storable]
-    public SymbolString BestTrainingSentence { get; set; }     // Currently set in RSquaredEvaluator: quite hacky, but makes testing much easier for now...
+    public SymbolList BestTrainingSentence { get; set; }     // Currently set in RSquaredEvaluator: quite hacky, but makes testing much easier for now...
     #endregion
 
@@ -185 +195 @@
       Parameters.Add(new FixedValueParameter<IntValue>(SearchDataStructureSizeParameterName, "The size of the search data structure.", new IntValue((int)1e5)));
       Parameters.Add(new FixedValueParameter<EnumValue<StorageType>>(SearchDataStructureParameterName, new EnumValue<StorageType>(StorageType.SortedSet)));
+      Parameters.Add(new FixedValueParameter<IntValue>(ConstantOptimizationIterationsParameterName, new IntValue(10)));
 
       SearchDataStructureParameter.Value.ValueChanged += (o, e) => Prepare();
@@ -240 +251 @@
     #endregion
 
-    [Storable]
-    private Dictionary<VariableTerminalSymbol, double> variableImportance;
-
     public override void Prepare() {
       DistinctSentencesComplexity = new Dictionary<int, int>();
@@ -259 +267 @@
       if (previousExecutionState != ExecutionState.Paused) {
         InitResults();
-        var phrase0 = new SymbolString(new[] { Grammar.StartSymbol });
+        var phrase0 = new SymbolList(new[] { Grammar.StartSymbol });
         var phrase0Hash = Grammar.Hasher.CalcHashCode(phrase0);
 
@@ -268 +276 @@
       var errorWeight = ErrorWeight;
       var optimizeConstants = OptimizeConstants; // cache value to avoid parameter lookup
+      var iterations = ConstantOptimizationIterations;
       // main search loop
       while (OpenPhrases.Count > 0) {
@@ -278 +287 @@
           continue;
 
-        SymbolString currPhrase = fetchedSearchNode.SymbolString;
+        SymbolList currPhrase = fetchedSearchNode.SymbolList;
 
         OnPhraseFetched(fetchedSearchNode.Hash, currPhrase);
@@ -292 +301 @@
           PhraseExpansionCount++;
 
-          SymbolString newPhrase = currPhrase.DerivePhrase(nonterminalSymbolIndex, appliedProductions[i]);
-          int newPhraseComplexity = Grammar.GetComplexity(newPhrase);
+          SymbolList newPhrase = currPhrase.DerivePhrase(nonterminalSymbolIndex, appliedProductions[i]);
+          int newPhraseComplexity = newPhrase.Complexity;
 
           if (newPhraseComplexity > MaxComplexity)
@@ -300 +309 @@
           var phraseHash = Grammar.Hasher.CalcHashCode(newPhrase);
 
-          OnPhraseDerived(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+          OnPhraseDerived(fetchedSearchNode.Hash, fetchedSearchNode.SymbolList, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
 
           if (newPhrase.IsSentence()) {
             AllGeneratedSentencesCount++;
 
-            OnSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+            OnSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolList, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
 
             // Is the best solution found? (only if RSquaredEvaluator is activated)
@@ -320 +329 @@
 
               DistinctSentencesComplexity[phraseHash] = newPhraseComplexity;
-              OnDistinctSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+              OnDistinctSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolList, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
             }
             UpdateView();
@@ -327 +336 @@
 
             bool isCompleteSentence = IsCompleteSentence(newPhrase);
-            double r2 = isCompleteSentence ? Grammar.EvaluatePhrase(newPhrase, Problem.ProblemData, optimizeConstants) : fetchedSearchNode.R2;
-            double phrasePriority = GetPriority(newPhrase, r2, MaxSentenceLength);
+            double r2 = isCompleteSentence ? Grammar.EvaluatePhrase(newPhrase, Problem.ProblemData, optimizeConstants, iterations) : fetchedSearchNode.R2;
+            double phrasePriority = GetPriority(newPhrase, r2);
 
             SearchNode newSearchNode = new SearchNode(phraseHash, phrasePriority, r2, newPhrase);
@@ -338 +347 @@
     }
 
-    protected static double GetPriority(SymbolString phrase, double r2, int maxSentenceLength) {
-      return (1 - r2) * phrase.Count();
-    }
-
-    private bool IsCompleteSentence(SymbolString phrase) {
+    protected static double GetPriority(SymbolList phrase, double r2) {
+      return (1 - r2) * phrase.Count;
+    }
+
+    private bool IsCompleteSentence(SymbolList phrase) {
       return !phrase.Any(x => x is NonterminalSymbol && x != Grammar.Expr);
     }
@@ -386 +395 @@
         Problem.ProblemData.TrainingIndices,
         applyLinearScaling: true,
-        maxIterations: 10,
+        maxIterations: ConstantOptimizationIterations,
         updateVariableWeights: false,
         updateConstantsInTree: true);
@@ -395 +404 @@
       Results.AddOrUpdateResult(BestTrainingModelResultName, model);
       Results.AddOrUpdateResult(BestTrainingSolutionResultName, bestTrainingSolution);
-      Results.AddOrUpdateResult(BestComplexityResultName, new IntValue(Grammar.GetComplexity(BestTrainingSentence)));
+      Results.AddOrUpdateResult(BestComplexityResultName, new IntValue(BestTrainingSentence.Complexity));
       base.OnStopped();
     }
@@ -450 +459 @@
 
     public event EventHandler<PhraseEventArgs> PhraseFetched;
-    private void OnPhraseFetched(int hash, SymbolString symbolString) {
+    private void OnPhraseFetched(int hash, SymbolList phrase) {
       if (PhraseFetched != null) {
-        PhraseFetched(this, new PhraseEventArgs(hash, symbolString));
+        PhraseFetched(this, new PhraseEventArgs(hash, phrase));
       }
     }
 
     public event EventHandler<PhraseAddedEventArgs> PhraseDerived;
-    private void OnPhraseDerived(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
+    private void OnPhraseDerived(int parentHash, SymbolList parentSymbolList, int addedHash, SymbolList addedSymbolList, Symbol expandedSymbol, SymbolList expandedProduction) {
       if (PhraseDerived != null) {
-        PhraseDerived(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
+        PhraseDerived(this, new PhraseAddedEventArgs(parentHash, parentSymbolList, addedHash, addedSymbolList, expandedSymbol, expandedProduction));
       }
     }
 
     public event EventHandler<PhraseAddedEventArgs> SentenceGenerated;
-    private void OnSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
+    private void OnSentenceGenerated(int parentHash, SymbolList parentSymbolList, int addedHash, SymbolList addedSymbolList, Symbol expandedSymbol, SymbolList expandedProduction) {
       if (SentenceGenerated != null) {
-        SentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
+        SentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolList, addedHash, addedSymbolList, expandedSymbol, expandedProduction));
       }
     }
 
     public event EventHandler<PhraseAddedEventArgs> DistinctSentenceGenerated;
-    private void OnDistinctSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
+    private void OnDistinctSentenceGenerated(int parentHash, SymbolList parentSymbolList, int addedHash, SymbolList addedSymbolList, Symbol expandedSymbol, SymbolList expandedProduction) {
       if (DistinctSentenceGenerated != null) {
-        DistinctSentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
+        DistinctSentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolList, addedHash, addedSymbolList, expandedSymbol, expandedProduction));
       }
     }
@@ -486 +495 @@
     public int Hash { get; }
 
-    public SymbolString Phrase { get; }
-
-    public PhraseEventArgs(int hash, SymbolString phrase) {
+    public SymbolList Phrase { get; }
+
+    public PhraseEventArgs(int hash, SymbolList phrase) {
       Hash = hash;
       Phrase = phrase;
@@ -498 +507 @@
     public int NewHash { get; }
 
-    public SymbolString ParentPhrase { get; }
-    public SymbolString NewPhrase { get; }
+    public SymbolList ParentPhrase { get; }
+    public SymbolList NewPhrase { get; }
 
     public Symbol ExpandedSymbol { get; }
 
-    public Production ExpandedProduction { get; }
-
-    public PhraseAddedEventArgs(int parentHash, SymbolString parentPhrase, int newHash, SymbolString newPhrase, Symbol expandedSymbol, Production expandedProduction) {
+    public SymbolList ExpandedProduction { get; }
+
+    public PhraseAddedEventArgs(int parentHash, SymbolList parentPhrase, int newHash, SymbolList newPhrase, Symbol expandedSymbol, SymbolList expandedProduction) {
       ParentHash = parentHash;
       ParentPhrase = parentPhrase;