Examples with NeuralNetwork org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork used on opensource projects

Example 1 with NeuralNetwork

use of org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork in project knime-core by knime.

the class PMMLNeuralNetworkTranslator method initializeFrom.

/**
 * {@inheritDoc}
 */
@Override
public void initializeFrom(final PMMLDocument pmmlDoc) {
    m_nameMapper = new DerivedFieldMapper(pmmlDoc);
    NeuralNetwork[] models = pmmlDoc.getPMML().getNeuralNetworkArray();
    if (models.length == 0) {
        throw new IllegalArgumentException("No neural network model" + " provided.");
    } else if (models.length > 1) {
        LOGGER.warn("Multiple neural network models found. " + "Only the first model is considered.");
    }
    NeuralNetwork nnModel = models[0];
    // ------------------------------
    // initiate Neural Input
    initInputLayer(nnModel);
    // -------------------------------
    // initiate Hidden Layer
    initiateHiddenLayers(nnModel);
    // -------------------------------
    // initiate Final Layer
    initiateFinalLayer(nnModel);
    // --------------------------------
    // initiate Neural Outputs
    initiateNeuralOutputs(nnModel);
    // --------------------------------
    // initiate Neural Network properties
    ACTIVATIONFUNCTION.Enum actFunc = nnModel.getActivationFunction();
    NNNORMALIZATIONMETHOD.Enum normMethod = nnModel.getNormalizationMethod();
    if (ACTIVATIONFUNCTION.LOGISTIC != actFunc) {
        LOGGER.error("Only logistic activation function is " + "supported in KNIME MLP.");
    }
    if (NNNORMALIZATIONMETHOD.NONE != normMethod) {
        LOGGER.error("No normalization method is " + "supported in KNIME MLP.");
    }
    MININGFUNCTION.Enum functionName = nnModel.getFunctionName();
    if (MININGFUNCTION.CLASSIFICATION == functionName) {
        m_mlpMethod = MultiLayerPerceptron.CLASSIFICATION_MODE;
    } else if (MININGFUNCTION.REGRESSION == functionName) {
        m_mlpMethod = MultiLayerPerceptron.REGRESSION_MODE;
    }
    if (m_allLayers.size() < 3) {
        throw new IllegalArgumentException("Only neural networks with 3 Layers supported in KNIME MLP.");
    }
    Layer[] allLayers = new Layer[m_allLayers.size()];
    allLayers = m_allLayers.toArray(allLayers);
    m_mlp = new MultiLayerPerceptron(allLayers);
    Architecture myarch = new Architecture(allLayers[0].getPerceptrons().length, allLayers.length - 2, allLayers[1].getPerceptrons().length, allLayers[allLayers.length - 1].getPerceptrons().length);
    m_mlp.setArchitecture(myarch);
    m_mlp.setClassMapping(m_classmap);
    m_mlp.setInputMapping(m_inputmap);
    m_mlp.setMode(m_mlpMethod);
}

Example 2 with NeuralNetwork

use of org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork in project knime-core by knime.

the class PMMLNeuralNetworkTranslator method exportTo.

/**
 * {@inheritDoc}
 */
@Override
public SchemaType exportTo(final PMMLDocument pmmlDoc, final PMMLPortObjectSpec spec) {
    m_nameMapper = new DerivedFieldMapper(pmmlDoc);
    NeuralNetwork nnModel = pmmlDoc.getPMML().addNewNeuralNetwork();
    PMMLMiningSchemaTranslator.writeMiningSchema(spec, nnModel);
    if (m_mlp.getMode() == MultiLayerPerceptron.CLASSIFICATION_MODE) {
        nnModel.setFunctionName(MININGFUNCTION.CLASSIFICATION);
    } else if (m_mlp.getMode() == MultiLayerPerceptron.REGRESSION_MODE) {
        nnModel.setFunctionName(MININGFUNCTION.REGRESSION);
    }
    nnModel.setAlgorithmName("RProp");
    nnModel.setActivationFunction(ACTIVATIONFUNCTION.LOGISTIC);
    nnModel.setNormalizationMethod(NNNORMALIZATIONMETHOD.NONE);
    nnModel.setWidth(0.0);
    nnModel.setNumberOfLayers(BigInteger.valueOf(m_mlp.getNrLayers() - 1));
    // add input layer
    addInputLayer(nnModel, m_mlp);
    // add hidden & final layers
    for (int i = 1; i < m_mlp.getNrLayers(); i++) {
        addLayer(nnModel, m_mlp, i);
    }
    // add output layer
    addOutputLayer(nnModel, m_mlp, spec);
    return NeuralNetwork.type;
}

Example 3 with NeuralNetwork

use of org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork in project knime-core by knime.

the class PMMLModelWrapper method getSegmentContent.

/**
 * Returns the content of a segment as a model wrapper.
 * @param s The segment
 * @return Returns a wrapper around the model
 */
public static PMMLModelWrapper getSegmentContent(final Segment s) {
    TreeModel treemodel = s.getTreeModel();
    if (treemodel != null) {
        return new PMMLTreeModelWrapper(treemodel);
    }
    RegressionModel regrmodel = s.getRegressionModel();
    if (regrmodel != null) {
        return new PMMLRegressionModelWrapper(regrmodel);
    }
    GeneralRegressionModel genregrmodel = s.getGeneralRegressionModel();
    if (genregrmodel != null) {
        return new PMMLGeneralRegressionModelWrapper(genregrmodel);
    }
    ClusteringModel clustmodel = s.getClusteringModel();
    if (clustmodel != null) {
        return new PMMLClusteringModelWrapper(clustmodel);
    }
    NaiveBayesModel nbmodel = s.getNaiveBayesModel();
    if (nbmodel != null) {
        return new PMMLNaiveBayesModelWrapper(nbmodel);
    }
    NeuralNetwork nn = s.getNeuralNetwork();
    if (nn != null) {
        return new PMMLNeuralNetworkWrapper(nn);
    }
    RuleSetModel rsmodel = s.getRuleSetModel();
    if (rsmodel != null) {
        return new PMMLRuleSetModelWrapper(rsmodel);
    }
    SupportVectorMachineModel svmmodel = s.getSupportVectorMachineModel();
    if (svmmodel != null) {
        return new PMMLSupportVectorMachineModelWrapper(svmmodel);
    }
    return null;
}

Example 4 with NeuralNetwork

use of org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork in project knime-core by knime.

the class PMMLPortObject method moveGlobalTransformationsToModel.

/**
 * Moves the content of the transformation dictionary to local
 * transformations of the model if a model exists.
 */
public void moveGlobalTransformationsToModel() {
    PMML pmml = m_pmmlDoc.getPMML();
    TransformationDictionary transDict = pmml.getTransformationDictionary();
    if (transDict == null || transDict.getDerivedFieldArray() == null || transDict.getDerivedFieldArray().length == 0) {
        // nothing to be moved
        return;
    }
    DerivedField[] globalDerivedFields = transDict.getDerivedFieldArray();
    LocalTransformations localTrans = null;
    if (pmml.getTreeModelArray().length > 0) {
        TreeModel model = pmml.getTreeModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.getClusteringModelArray().length > 0) {
        ClusteringModel model = pmml.getClusteringModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.getNeuralNetworkArray().length > 0) {
        NeuralNetwork model = pmml.getNeuralNetworkArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.getSupportVectorMachineModelArray().length > 0) {
        SupportVectorMachineModel model = pmml.getSupportVectorMachineModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.getRegressionModelArray().length > 0) {
        RegressionModel model = pmml.getRegressionModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.getGeneralRegressionModelArray().length > 0) {
        GeneralRegressionModel model = pmml.getGeneralRegressionModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    } else if (pmml.sizeOfRuleSetModelArray() > 0) {
        RuleSetModel model = pmml.getRuleSetModelArray(0);
        localTrans = model.getLocalTransformations();
        if (localTrans == null) {
            localTrans = model.addNewLocalTransformations();
        }
    }
    if (localTrans != null) {
        DerivedField[] derivedFields = appendDerivedFields(localTrans.getDerivedFieldArray(), globalDerivedFields);
        localTrans.setDerivedFieldArray(derivedFields);
        // remove derived fields from TransformationDictionary
        transDict.setDerivedFieldArray(new DerivedField[0]);
    }
// else do nothing as no model exists yet
}

Example 5 with NeuralNetwork

use of org.dmg.pmml.NeuralNetworkDocument.NeuralNetwork in project knime-core by knime.

the class PMMLNeuralNetworkTranslator method exportTo.

/**
 * {@inheritDoc}
 */
@Override
public SchemaType exportTo(final PMMLDocument pmmlDoc, final PMMLPortObjectSpec spec) {
    m_nameMapper = new DerivedFieldMapper(pmmlDoc);
    NeuralNetwork nnModel = pmmlDoc.getPMML().addNewNeuralNetwork();
    PMMLMiningSchemaTranslator.writeMiningSchema(spec, nnModel);
    if (m_mlp.getMode() == MultiLayerPerceptron.CLASSIFICATION_MODE) {
        nnModel.setFunctionName(MININGFUNCTION.CLASSIFICATION);
    } else if (m_mlp.getMode() == MultiLayerPerceptron.REGRESSION_MODE) {
        nnModel.setFunctionName(MININGFUNCTION.REGRESSION);
    }
    nnModel.setAlgorithmName("RProp");
    nnModel.setActivationFunction(ACTIVATIONFUNCTION.LOGISTIC);
    nnModel.setNormalizationMethod(NNNORMALIZATIONMETHOD.NONE);
    nnModel.setWidth(0.0);
    nnModel.setNumberOfLayers(BigInteger.valueOf(m_mlp.getNrLayers() - 1));
    // add input layer
    addInputLayer(nnModel, m_mlp);
    // add hidden & final layers
    for (int i = 1; i < m_mlp.getNrLayers(); i++) {
        addLayer(nnModel, m_mlp, i);
    }
    // add output layer
    addOutputLayer(nnModel, m_mlp, spec);
    return NeuralNetwork.type;
}