Package statechum.analysis.learning.MarkovModel

Examples of statechum.analysis.learning.MarkovModel.MarkovOutcome

649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
        for(Label label:graph.getCache().getAlphabet())
        {
          if (!failureLabels.contains(label))
          {// if the labels is not already recorded as being inconsistently predicted
            MarkovOutcome predictedFromEalierTrace = outgoing_labels_probabilities.get(label);
              Trace Predicted_trace = new Trace();
              if (predictionGraphInverted)
              {
                for(int i=pathToNewState.size()-1;i>=0;--i) Predicted_trace.add(pathToNewState.get(i));if (pathBeyondCurrentState != null) Predicted_trace.getList().addAll(pathBeyondCurrentState);
              }
              else
              {
                Predicted_trace.getList().addAll(pathToNewState);
              }
              Predicted_trace.add(label);
             
              MarkovOutcome predicted_from_Markov=model.predictionsMatrix.get(Predicted_trace);
            MarkovOutcome outcome = MarkovOutcome.reconcileOpinions_PosNeg_Overrides_Null(predictedFromEalierTrace, predicted_from_Markov);
            if (outcome != predictedFromEalierTrace)
            {// we learnt something new, be it a new value (or a non-null value) or a failure, record it
              if (outcome == MarkovOutcome.failure)
              {
                failureLabels.add(label);outgoing_labels_probabilities.remove(label);
View Full Code Here
715
716
717
718
719
720
721
722
723
724
725
          {
            predictedTrace.getList().addAll(pathToUseWithMarkovToPredictOutgoing);
          }
          predictedTrace.add(lbl);
         
          MarkovOutcome newValue = null;
          UpdatablePairInteger p=model.occurrenceMatrix.get(predictedTrace);if (p == null) { p=new UpdatablePairInteger(0, 0);model.occurrenceMatrix.put(predictedTrace,p); }

          boolean foundAccept = false, foundReject = false;
          for(Object vObj:graphToCheckForConsistency.getTargets(targets))
          {
View Full Code Here
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
        }
        //System.out.println(vert.toString()+" : "+encounteredPartOfTrace+" outgoing: "+outgoingLabels);
        if (model.occurrenceMatrix.containsKey(new Trace(encounteredPartOfTrace, true))) // we skip everything where a path was not seen in PTA.
            for(Label label:outgoingLabels)
            {
            MarkovOutcome labels_occurrence= outgoing_labels_value.get(label);
            if (labels_occurrence != MarkovOutcome.failure)
            {
                Trace traceToCheck = new Trace();traceToCheck.getList().addAll(encounteredPartOfTrace);
                traceToCheck.add(label);

                MarkovOutcome predicted_from_Markov=model.predictionsMatrix.get(traceToCheck);
                if (predicted_from_Markov != MarkovOutcome.failure)
                {// if training data does not lead to a consistent outcome for this label because chunk length is too small, not much we can do, but otherwise we are here and can make use of the data
                  if (!checker.consistent(labels_occurrence, predicted_from_Markov))
                  {
                    inconsistencies.addAndGet(1);// record inconsistency
View Full Code Here
160
161
162
163
164
165
166
167
168
169
170
    }
    for(Entry<Label,CmpVertex> state:states.entrySet())
    {
      for(Label label:ptaClassifier.graph.getCache().getAlphabet())
      {
        MarkovOutcome transition = ptaClassifier.model.predictionsMatrix.get(new Trace(Arrays.asList(new Label[]{state.getKey(),label}),true));
        if (transition != null)
          if (transition == MarkovOutcome.positive) outcome.transitionMatrix.get(state.getValue()).put(label,states.get(label));
      }
    }   
View Full Code Here
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
    long scoreCurrentFanout = 0, score = 0;
    Map<Label,MarkovOutcome> outgoing_red_probabilities=cl.predictTransitionsFromState(red,pathLenBeyondCurrentState,cl.model.getChunkLen(),null);
    Map<Label,MarkovOutcome> outgoing_blue_probabilities=cl.predictTransitionsFromState(blue,pathLenBeyondCurrentState,cl.model.getChunkLen(),null);
    for(Entry<Label,MarkovOutcome> entry:outgoing_red_probabilities.entrySet())
    {
      MarkovOutcome outcomeBlue = outgoing_blue_probabilities.get(entry.getKey());
      if (outcomeBlue == null && entry.getValue() == MarkovOutcome.negative)
        ++scoreCurrentFanout; // red negative, blue absent, hence the two are consistent
      if (outcomeBlue == entry.getValue()) // or if the two are consistent
      {
        if (stepNumber > 1)
        {
          LinkedList<Label> pathBeyond = new LinkedList<Label>(pathLenBeyondCurrentState);pathBeyond.add(entry.getKey());
          score+=comparePredictedFanouts(cl,red,blue,pathBeyond,stepNumber-1);
        }
        ++scoreCurrentFanout;
      }
    }
     
    for(Entry<Label,MarkovOutcome> entry:outgoing_blue_probabilities.entrySet())
    {
      MarkovOutcome outcomeRed = outgoing_red_probabilities.get(entry.getKey());
      if (outcomeRed == null && entry.getValue() == MarkovOutcome.negative)
        ++scoreCurrentFanout; // blue negative, red absent, hence the two are consistent
      if (outcomeRed == entry.getValue()) // or if the two are consistent
      {
        if (stepNumber > 1)
View Full Code Here
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
        for(Label label:graph.getCache().getAlphabet())
        {
          if (!failureLabels.contains(label))
          {// if the labels is not already recorded as being inconsistently predicted
            MarkovOutcome predictedFromEalierTrace = outgoing_labels_probabilities.get(label);
             
              PredictionForSequence prediction = MarkovMatrixEngine.getPredictionIfExists(lastElementToPrediction, label);
              MarkovOutcome predicted_from_Markov= prediction!=null?prediction.prediction:null;
            MarkovOutcome outcome = MarkovOutcome.reconcileOpinions_PosNeg_Overrides_Null(predictedFromEalierTrace, predicted_from_Markov);
            if (outcome != predictedFromEalierTrace)
            {// we learnt something new, be it a new value (or a non-null value) or a failure, record it
              if (outcome == MarkovOutcome.failure)
              {
                failureLabels.add(label);outgoing_labels_probabilities.remove(label);
View Full Code Here
897
898
899
900
901
902
903
904
905
906
907
            pathToUpdateInMarkov.addAll(pathToUseWithMarkovToPredictOutgoing);
          }
         
          pathToUpdateInMarkov.add(lbl);
         
          MarkovOutcome newValue = null;
          PredictionForSequence prediction = model.markovMatrix.getPredictionAndCreateNewOneIfNecessary(pathToUpdateInMarkov);
         
          boolean foundAccept = false, foundReject = false;
          for(Object vObj:graphToCheckForConsistency.getTargets(targets))
          {
View Full Code Here
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
       
        //System.out.println(vert.toString()+" : "+encounteredPartOfTrace+" outgoing: "+outgoingLabels);
        if (checker.considerPathsWithPrefixMissingInMarkov() || mapFromLastLabelToNodes != null) // we skip everything where a path was not seen in PTA unless we are asked to consider all such paths.
            for(Label label:outgoingLabels)
            {
            MarkovOutcome labels_occurrence= outgoing_labels_value.get(label);
            if (labels_occurrence != MarkovOutcome.failure)
            {
                PredictionForSequence prediction = MarkovMatrixEngine.getPredictionIfExists(mapFromLastLabelToNodes, label);
                MarkovOutcome predicted_from_Markov=prediction == null?null:prediction.prediction;
                if (predicted_from_Markov != MarkovOutcome.failure)
                {// if training data does not lead to a consistent outcome for this label because chunk length is too small, not much we can do, but otherwise we are here and can make use of the data
                  if (!checker.consistent(labels_occurrence, predicted_from_Markov))
                  {
                    inconsistencies.addAndGet(1);// record inconsistency
View Full Code Here
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
        }
       
       
        for(Entry<Label,CmpVertex> existing:entry.getValue().entrySet())
        {
          MarkovOutcome predictedTarget = predictions.get(existing.getKey());
         
          if (existing.getValue().isAccept() && predictedTarget == MarkovOutcome.positive)
            ++numberOfExistingPredicted;
          if (!existing.getValue().isAccept() && predictedTarget == MarkovOutcome.negative)
            ++numberOfExistingPredicted;
View Full Code Here
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
    long scoreCurrentFanout = 0, score = 0;
    Map<Label,MarkovOutcome> outgoing_red_probabilities=cl.predictTransitionsFromState(red,pathLenBeyondCurrentState,cl.model.getChunkLen(),null);
    Map<Label,MarkovOutcome> outgoing_blue_probabilities=cl.predictTransitionsFromState(blue,pathLenBeyondCurrentState,cl.model.getChunkLen(),null);
    for(Entry<Label,MarkovOutcome> entry:outgoing_red_probabilities.entrySet())
    {
      MarkovOutcome outcomeBlue = outgoing_blue_probabilities.get(entry.getKey());
      if (outcomeBlue == null && entry.getValue() == MarkovOutcome.negative)
        ++scoreCurrentFanout; // red negative, blue absent, hence the two are consistent
      if (outcomeBlue == entry.getValue()) // or if the two are consistent
      {
        if (stepNumber > 1)
        {
          LinkedList<Label> pathBeyond = new LinkedList<Label>(pathLenBeyondCurrentState);pathBeyond.add(entry.getKey());
          score+=comparePredictedFanouts(cl,red,blue,pathBeyond,stepNumber-1);
        }
        ++scoreCurrentFanout;
      }
    }
     
    for(Entry<Label,MarkovOutcome> entry:outgoing_blue_probabilities.entrySet())
    {
      MarkovOutcome outcomeRed = outgoing_red_probabilities.get(entry.getKey());
      if (outcomeRed == null && entry.getValue() == MarkovOutcome.negative)
        ++scoreCurrentFanout; // blue negative, red absent, hence the two are consistent
      if (outcomeRed == entry.getValue()) // or if the two are consistent
      {
        if (stepNumber > 1)
View Full Code Here
TOP

Related Classes of statechum.analysis.learning.MarkovModel.MarkovOutcome

  • statechum.analysis.learning.experiments.PairSelection.MarkovPassivePairSelection
  • statechum.analysis.learning.experiments.PairSelection.MarkovScoreComputation
  • statechum.analysis.learning.MarkovClassifier
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.