Package org.encog.ml.ea.train.basic

Examples of org.encog.ml.ea.train.basic.TrainEA

49
50
51
52
53
54
55
56
57
   */
  public MLTrain create(final MLMethod method,
      final MLDataSet training, final String argsStr) {

    final CalculateScore score = new TrainingSetScore(training);   
    final TrainEA train = NEATUtil.constructNEATTrainer((NEATPopulation)method, score);

    return train;
  }
View Full Code Here
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
   * @param calculateScore The score function.
   * @return The NEAT EA trainer.
   */
  public static TrainEA constructNEATTrainer(final NEATPopulation population,
      final CalculateScore calculateScore) {
    final TrainEA result = new TrainEA(population, calculateScore);
    result.setSpeciation(new OriginalNEATSpeciation());

    result.setSelection(new TruncationSelection(result, 0.3));
    final CompoundOperator weightMutation = new CompoundOperator();
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(1),
            new MutatePerturbLinkWeight(0.02)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(2),
            new MutatePerturbLinkWeight(0.02)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(3),
            new MutatePerturbLinkWeight(0.02)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectProportion(0.02),
            new MutatePerturbLinkWeight(0.02)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(1),
            new MutatePerturbLinkWeight(1)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(2),
            new MutatePerturbLinkWeight(1)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectFixed(3),
            new MutatePerturbLinkWeight(1)));
    weightMutation.getComponents().add(
        0.1125,
        new NEATMutateWeights(new SelectProportion(0.02),
            new MutatePerturbLinkWeight(1)));
    weightMutation.getComponents().add(
        0.03,
        new NEATMutateWeights(new SelectFixed(1),
            new MutateResetLinkWeight()));
    weightMutation.getComponents().add(
        0.03,
        new NEATMutateWeights(new SelectFixed(2),
            new MutateResetLinkWeight()));
    weightMutation.getComponents().add(
        0.03,
        new NEATMutateWeights(new SelectFixed(3),
            new MutateResetLinkWeight()));
    weightMutation.getComponents().add(
        0.01,
        new NEATMutateWeights(new SelectProportion(0.02),
            new MutateResetLinkWeight()));
    weightMutation.getComponents().finalizeStructure();

    result.setChampMutation(weightMutation);
    result.addOperation(0.5, new NEATCrossover());
    result.addOperation(0.494, weightMutation);
    result.addOperation(0.0005, new NEATMutateAddNode());
    result.addOperation(0.005, new NEATMutateAddLink());
    result.addOperation(0.0005, new NEATMutateRemoveLink());
    result.getOperators().finalizeStructure();

    if (population.isHyperNEAT()) {
      result.setCODEC(new HyperNEATCODEC());
    } else {
      result.setCODEC(new NEATCODEC());
    }

    return result;
  }
View Full Code Here
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
      final MLDataSet training, final String argsStr) {
   
    PrgPopulation pop = (PrgPopulation)method;
   
    final CalculateScore score = new TrainingSetScore(training);   
    TrainEA train = new TrainEA(pop, score);
    train.getRules().addRewriteRule(new RewriteConstants());
    train.getRules().addRewriteRule(new RewriteAlgebraic());
    train.setCODEC(new PrgCODEC());
    train.addOperation(0.8, new SubtreeCrossover());
    train.addOperation(0.1, new SubtreeMutation(pop.getContext(),4));
    train.addOperation(0.1, new ConstMutation(pop.getContext(),0.5,1.0));
    train.addScoreAdjuster(new ComplexityAdjustedScore());
    train.setSpeciation(new PrgSpeciation());
    return train;
  }
View Full Code Here
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
    EncogProgramContext context = new EncogProgramContext();
    StandardExtensions.createNumericOperators(context);
    PrgPopulation pop = new PrgPopulation(context,1);
    CalculateScore score = new ZeroEvalScoreFunction();

    TrainEA genetic = new TrainEA(pop, score);
    genetic.setValidationMode(true);
    genetic.setCODEC(new PrgCODEC());
    genetic.addOperation(0.95, new SubtreeCrossover());
    genetic.addOperation(0.05, new SubtreeMutation(context,4));
    genetic.addScoreAdjuster(new ComplexityAdjustedScore());
    genetic.getRules().addRewriteRule(new RewriteConstants());
    genetic.getRules().addRewriteRule(new RewriteAlgebraic());

    EncogProgram expression = new EncogProgram(context);
    expression.compileExpression(start);
    RenderCommonExpression render = new RenderCommonExpression();
    genetic.getRules().rewrite(expression);
    Assert.assertEquals(expect, render.render(expression));
  }
View Full Code Here
TOP

Related Classes of org.encog.ml.ea.train.basic.TrainEA

  • org.encog.ml.factory.train.EPLGAFactory
  • org.encog.ml.factory.train.NEATGAFactory
  • org.encog.ml.prg.train.rewrite.TestRewriteAlgebraic
  • org.encog.ml.train.strategy.end.EndTrainingStrategy
  • org.encog.neural.neat.NEATUtil
  • org.encog.neural.networks.training.TrainingSetScore
    • 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.