Package org.tribuo.multilabel.example

Class MultiLabelGaussianDataSource

java.lang.Object
org.tribuo.multilabel.example.MultiLabelGaussianDataSource
All Implemented Interfaces:
com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<DataSourceProvenance>, Iterable<Example<MultiLabel>>, ConfigurableDataSource<MultiLabel>, DataSource<MultiLabel>
public final class MultiLabelGaussianDataSource extends Object implements ConfigurableDataSource<MultiLabel>
Generates a multi label output drawn from a series of functions.

The functions are:

  • y_0 is positive if N(w_00*x_0 + w_01*x_1 + w_02*x_1*x_0 + w_03*x_1*x_1*x_1,variance) > threshold_0.
  • y_1 is positive if N(w_10*x_0 + w_11*x_1 + w_12*x_1*x_0 + w_13*x_1*x_1*x_1,variance) < threshold_1.
  • y_2 is positive if N(w_20*x_0 + w_21*x_2 + w_22*x_1*x_0 + w_23*x_1*x_2*x_2,variance) > threshold_2.
There are 4 features x_0, x_1, x_2, x_3. x_0 and x_1 are used by all three labels, x_2 is only used by y_2, and x_3 is irrelevant.

By default y_1 is the inverse of y_0, and y_2 shares the same weights w_0 and w_2.

  • y_0 weights = [1.0,1.0,1.0,1.0]
  • y_1 weights = [1.0,1.0,1.0,1.0]
  • y_2 weights = [1.0,-3.0,1.0,3.0]
  • threshold = [0.0,0.0,2.0

The features are drawn from a uniform distribution over the range.

  • Constructor Details

    • MultiLabelGaussianDataSource

      public MultiLabelGaussianDataSource(int numSamples, float[] yZeroWeights, float[] yOneWeights, float[] yTwoWeights, float[] threshold, boolean[] negate, float variance, float[] xMin, float[] xMax, long seed)
      Generates a multi-label output drawn from three gaussian functions.
      • N(w_00*x_0 + w_01*x_1 + w_02*x_1*x_0 + w_03*x_1*x_1*x_1,variance)
      • N(w_10*x_0 + w_11*x_1 + w_12*x_1*x_0 + w_13*x_1*x_1*x_1,variance)
      • N(w_20*x_0 + w_21*x_2 + w_22*x_1*x_0 + w_23*x_1*x_2*x_2,variance)

      The features are drawn from a uniform distribution over the range.

      Parameters:
      numSamples - The size of the output dataset.
      yZeroWeights - The feature weights for label y_0.
      yOneWeights - The feature weights for label y_1.
      yTwoWeights - The feature weights for label y_2.
      threshold - The y threshold of each label.
      negate - Should the computed value be negated before thresholding?
      variance - The variance of the gaussian.
      xMin - The minimum feature values (inclusive).
      xMax - The maximum feature values (exclusive).
      seed - The rng seed to use.

  • Method Details

    • postConfig

      public void postConfig()
      Used by the OLCUT configuration system, and should not be called by external code.
      Specified by:
      postConfig in interface com.oracle.labs.mlrg.olcut.config.Configurable

    • getOutputFactory

      public OutputFactory<MultiLabel> getOutputFactory()
      Description copied from interface: DataSource
      Returns the OutputFactory associated with this Output subclass.
      Specified by:
      getOutputFactory in interface DataSource<MultiLabel>
      Returns:
      The output factory.

    • getProvenance

      public DataSourceProvenance getProvenance()
      Specified by:
      getProvenance in interface com.oracle.labs.mlrg.olcut.provenance.Provenancable<DataSourceProvenance>

    • iterator

      public Iterator<Example<MultiLabel>> iterator()
      Specified by:
      iterator in interface Iterable<Example<MultiLabel>>

    • generateDataset

      public static MutableDataset<MultiLabel> generateDataset(int numSamples, float[] yZeroWeights, float[] yOneWeights, float[] yTwoWeights, float[] threshold, boolean[] negate, float variance, float[] xMin, float[] xMax, long seed)
      Generates a multi-label output drawn from three gaussian functions.
      • N(w_00*x_0 + w_01*x_1 + w_02*x_1*x_0 + w_03*x_1*x_1*x_1,variance)
      • N(w_10*x_0 + w_11*x_1 + w_12*x_1*x_0 + w_13*x_1*x_1*x_1,variance)
      • N(w_20*x_0 + w_21*x_2 + w_22*x_1*x_0 + w_23*x_1*x_2*x_2,variance)

      The features are drawn from a uniform distribution over the range.

      Parameters:
      numSamples - The size of the output dataset.
      yZeroWeights - The feature weights for label y_0.
      yOneWeights - The feature weights for label y_1.
      yTwoWeights - The feature weights for label y_2.
      threshold - The y threshold of each label.
      negate - Should the computed value be negated before thresholding?
      variance - The variance of the gaussian.
      xMin - The minimum feature values (inclusive).
      xMax - The maximum feature values (exclusive).
      seed - The rng seed to use.
      Returns:
      A dataset drawn from several gaussian generated labels.

    • makeDefaultSource

      public static MultiLabelGaussianDataSource makeDefaultSource(int numSamples, long seed)
      Generates a multi label output drawn from a series of functions.

      The functions are:

      • y_0 is positive if N(w_00*x_0 + w_01*x_1 + w_02*x_1*x_0 + w_03*x_1*x_1*x_1,variance) > threshold_0.
      • y_1 is positive if N(w_10*x_0 + w_11*x_1 + w_12*x_1*x_0 + w_13*x_1*x_1*x_1,variance) < threshold_1.
      • y_2 is positive if N(w_20*x_0 + w_21*x_2 + w_22*x_1*x_0 + w_23*x_1*x_2*x_2,variance) > threshold_2.
      There are 4 features x_0, x_1, x_2, x_3. x_0 and x_1 are used by all three labels, x_2 is only used by y_2, and x_3 is irrelevant.

      By default y_1 is the inverse of y_0, and y_2 shares the same weights w_0 and w_2.

      • y_0 weights = [1.0,1.0,1.0,1.0]
      • y_1 weights = [1.0,1.0,1.0,1.0]
      • y_2 weights = [1.0,-3.0,1.0,3.0]
      • threshold = [0.0,0.0,2.0

      The features are drawn from a uniform distribution over the range.

      Parameters:
      numSamples - The number of samples to draw.
      seed - The RNG seed.
      Returns:
      A dataset drawn from multiple Gaussians.