Uses of Interface
org.tribuo.math.la.SGDVector

Packages that use SGDVector

Package	Description
org.tribuo.classification.sgd	Provides infrastructure for Stochastic Gradient Descent for classification problems.
org.tribuo.classification.sgd.crf	Provides an implementation of a linear chain CRF trained using Stochastic Gradient Descent.
org.tribuo.classification.sgd.objectives	Provides classification loss functions for Stochastic Gradient Descent.
org.tribuo.common.sgd	Provides the base classes for models trained with stochastic gradient descent.
org.tribuo.interop.tensorflow	Provides an interface to TensorFlow, allowing the training of non-sequential models using any supported Tribuo output type.
org.tribuo.math	Contains the implementation of Tribuo's math library, it's gradient descent optimisers, kernels and a set of math related utils.
org.tribuo.math.la	Provides a linear algebra system used for numerical operations in Tribuo.
org.tribuo.math.optimisers.util	Provides some utility tensors for use in gradient optimisers.
org.tribuo.multilabel.sgd	Provides infrastructure for Stochastic Gradient Descent for multi-label classification problems.
org.tribuo.multilabel.sgd.linear	Provides an implementation of a multi-label classification linear model using Stochastic Gradient Descent.
org.tribuo.multilabel.sgd.objectives	Provides multi-label classification loss functions for Stochastic Gradient Descent.
org.tribuo.regression.sgd	Provides infrastructure for Stochastic Gradient Descent based regression models.
org.tribuo.regression.sgd.objectives	Provides regression loss functions for Stochastic Gradient Descent.

Uses of SGDVector in org.tribuo.classification.sgd

Fields in org.tribuo.classification.sgd declared as SGDVector

Modifier and Type	Field	Description
final SGDVector[][]	Util.SequenceExampleArray.features

Methods in org.tribuo.classification.sgd that return types with arguments of type SGDVector

Modifier and Type	Method	Description
default com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LabelObjective.lossAndGradient(Integer truth, SGDVector prediction)
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LabelObjective.valueAndGradient(int truth, SGDVector prediction)	Deprecated. In 4.1, to migrate to the new name LabelObjective.lossAndGradient(java.lang.Integer, org.tribuo.math.la.SGDVector).

Methods in org.tribuo.classification.sgd with parameters of type SGDVector

Modifier and Type	Method	Description
default com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LabelObjective.lossAndGradient(Integer truth, SGDVector prediction)
static Util.SequenceExampleArray	Util.shuffle(SGDVector[][] features, int[][] labels, double[] weights, SplittableRandom rng)	Shuffles a sequence of features, labels and weights, returning a tuple of the shuffled values.
static void	Util.shuffleInPlace(SGDVector[][] features, int[][] labels, double[] weights, SplittableRandom rng)	In place shuffle used for sequence problems.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LabelObjective.valueAndGradient(int truth, SGDVector prediction)	Deprecated. In 4.1, to migrate to the new name LabelObjective.lossAndGradient(java.lang.Integer, org.tribuo.math.la.SGDVector).

Uses of SGDVector in org.tribuo.classification.sgd.crf

Methods in org.tribuo.classification.sgd.crf that return SGDVector

Modifier and Type	Method	Description
static <T extends Output<T>> SGDVector[]	CRFModel.convertToVector(SequenceExample<T> example, ImmutableFeatureMap featureIDMap)	Converts a SequenceExample into an array of SGDVectors suitable for CRF prediction.

Methods in org.tribuo.classification.sgd.crf with parameters of type SGDVector

Modifier and Type	Method	Description
ChainHelper.ChainCliqueValues	CRFParameters.getCliqueValues(SGDVector[] features)	Generates the local scores and tuples them with the label - label transition weights.
DenseVector[]	CRFParameters.getLocalScores(SGDVector[] features)	Generate the local scores (i.e., the linear classifier for each token).
int[]	CRFParameters.predict(SGDVector[] features)	Generate a prediction using Viterbi.
List<Double>	CRFParameters.predictConfidenceUsingCBP(SGDVector[] features, List<Chunk> chunks)	This predicts per chunk confidence using the constrained forward backward algorithm from Culotta and McCallum 2004.
DenseVector[]	CRFParameters.predictMarginals(SGDVector[] features)	Generate a prediction using Belief Propagation.
com.oracle.labs.mlrg.olcut.util.Pair<Double,Tensor[]>	CRFParameters.valueAndGradient(SGDVector[] features, int[] labels)	Generates predictions based on the input features and labels, then scores those predictions to produce a loss for the example and a gradient update.

Uses of SGDVector in org.tribuo.classification.sgd.objectives

Methods in org.tribuo.classification.sgd.objectives that return types with arguments of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.lossAndGradient(Integer truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LogMulticlass.lossAndGradient(Integer truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.valueAndGradient(int truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LogMulticlass.valueAndGradient(int truth, SGDVector prediction)	Deprecated.

Methods in org.tribuo.classification.sgd.objectives with parameters of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.lossAndGradient(Integer truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LogMulticlass.lossAndGradient(Integer truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.valueAndGradient(int truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	LogMulticlass.valueAndGradient(int truth, SGDVector prediction)	Deprecated.

Uses of SGDVector in org.tribuo.common.sgd

Methods in org.tribuo.common.sgd that return types with arguments of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SGDObjective.lossAndGradient(T truth, SGDVector prediction)	Scores a prediction, returning the loss and a vector of per output dimension gradients.

Methods in org.tribuo.common.sgd with parameters of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SGDObjective.lossAndGradient(T truth, SGDVector prediction)	Scores a prediction, returning the loss and a vector of per output dimension gradients.
static <T> void	AbstractSGDTrainer.shuffleInPlace(SGDVector[] features, T[] labels, double[] weights, SplittableRandom rng)	Shuffles the features, outputs and weights in place.

Uses of SGDVector in org.tribuo.interop.tensorflow

Methods in org.tribuo.interop.tensorflow with parameters of type SGDVector

Modifier and Type	Method	Description
TensorMap	DenseFeatureConverter.convert(SGDVector vector)
TensorMap	FeatureConverter.convert(SGDVector vector)	Converts a SGDVector representing the features into a TensorMap.
TensorMap	ImageConverter.convert(SGDVector vector)

Method parameters in org.tribuo.interop.tensorflow with type arguments of type SGDVector

Modifier and Type	Method	Description
TensorMap	DenseFeatureConverter.convert(List<? extends SGDVector> vectors)
TensorMap	FeatureConverter.convert(List<? extends SGDVector> vectors)	Converts a list of SGDVectors representing a batch of features into a TensorMap.
TensorMap	ImageConverter.convert(List<? extends SGDVector> vectors)

Uses of SGDVector in org.tribuo.math

Methods in org.tribuo.math with parameters of type SGDVector

Modifier and Type	Method	Description
Tensor[]	FeedForwardParameters.gradients(com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector> score, SGDVector features)	Generates the parameter gradients given the loss, output gradient and input features.
Tensor[]	LinearParameters.gradients(com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector> score, SGDVector features)	Generate the gradients for a particular feature vector given the loss and the per output gradients.
DenseVector	FeedForwardParameters.predict(SGDVector example)	Generates an un-normalized prediction by feeding the features through the parameters.
DenseVector	LinearParameters.predict(SGDVector example)	Generates an unnormalised prediction by leftMultiply'ing the weights with the incoming features.

Method parameters in org.tribuo.math with type arguments of type SGDVector

Modifier and Type	Method	Description
Tensor[]	FeedForwardParameters.gradients(com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector> score, SGDVector features)	Generates the parameter gradients given the loss, output gradient and input features.
Tensor[]	LinearParameters.gradients(com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector> score, SGDVector features)	Generate the gradients for a particular feature vector given the loss and the per output gradients.

Uses of SGDVector in org.tribuo.math.la

Classes in org.tribuo.math.la that implement SGDVector

Modifier and Type	Class	Description
class	DenseVector	A dense vector, backed by a double array.
class	SparseVector	A sparse vector.

Methods in org.tribuo.math.la that return SGDVector

Modifier and Type	Method	Description
SGDVector	SGDVector.add(SGDVector other)	Adds other to this vector, producing a new SGDVector.
SGDVector	SparseVector.add(SGDVector other)	Adds other to this vector, producing a new SGDVector.
SGDVector	SGDVector.copy()	Returns a deep copy of this vector.
SGDVector	Matrix.getRow(int i)	Extract a row as an SGDVector.
SGDVector	Matrix.leftMultiply(SGDVector input)	Multiplies this Matrix by a SGDVector returning a vector of the appropriate size.
SGDVector	Matrix.rightMultiply(SGDVector input)	Multiplies this Matrix by a SGDVector returning a vector of the appropriate size.
SGDVector	SGDVector.scale(double coefficient)	Generates a new vector with each element scaled by coefficient.
SGDVector	SGDVector.subtract(SGDVector other)	Subtracts other from this vector, producing a new SGDVector.
SGDVector	SparseVector.subtract(SGDVector other)	Subtracts other from this vector, producing a new SGDVector.

Methods in org.tribuo.math.la with parameters of type SGDVector

Modifier and Type	Method	Description
DenseVector	DenseVector.add(SGDVector other)	Adds other to this vector, producing a new DenseVector.
SGDVector	SGDVector.add(SGDVector other)	Adds other to this vector, producing a new SGDVector.
SGDVector	SparseVector.add(SGDVector other)	Adds other to this vector, producing a new SGDVector.
void	DenseMatrix.broadcastIntersectAndAddInPlace(SGDVector input, boolean broadcastOverDim1)	Broadcasts the input vector and adds it to each row/column of the matrix.
default double	SGDVector.cosineDistance(SGDVector other)	Calculates the cosine distance of two vectors.
default double	SGDVector.cosineSimilarity(SGDVector other)	Calculates the cosine similarity of two vectors.
double	SparseVector.distance(SGDVector other, DoubleUnaryOperator transformFunc, DoubleUnaryOperator normalizeFunc)
double	DenseVector.dot(SGDVector other)
double	SGDVector.dot(SGDVector other)	Calculates the dot product between this vector and other.
double	SparseVector.dot(SGDVector other)
double	DenseVector.euclideanDistance(SGDVector other)	The l2 or euclidean distance between this vector and the other vector.
double	SGDVector.euclideanDistance(SGDVector other)	The l2 or euclidean distance between this vector and the other vector.
double	SparseVector.euclideanDistance(SGDVector other)
double	DenseVector.l1Distance(SGDVector other)	The l1 or Manhattan distance between this vector and the other vector.
double	SGDVector.l1Distance(SGDVector other)	The l1 or Manhattan distance between this vector and the other vector.
double	SparseVector.l1Distance(SGDVector other)
default double	SGDVector.l2Distance(SGDVector other)	Synonym for euclideanDistance.
DenseVector	DenseMatrix.leftMultiply(SGDVector input)
DenseVector	DenseSparseMatrix.leftMultiply(SGDVector input)
SGDVector	Matrix.leftMultiply(SGDVector input)	Multiplies this Matrix by a SGDVector returning a vector of the appropriate size.
Matrix	DenseVector.outer(SGDVector other)
Matrix	SGDVector.outer(SGDVector other)	Generates the matrix representing the outer product between the two vectors.
Matrix	SparseVector.outer(SGDVector other)	This generates the outer product when dotted with another SparseVector.
DenseVector	DenseMatrix.rightMultiply(SGDVector input)
DenseVector	DenseSparseMatrix.rightMultiply(SGDVector input)	rightMultiply is very inefficient on DenseSparseMatrix due to the storage format.
SGDVector	Matrix.rightMultiply(SGDVector input)	Multiplies this Matrix by a SGDVector returning a vector of the appropriate size.
DenseVector	DenseVector.subtract(SGDVector other)	Subtracts other from this vector, producing a new DenseVector.
SGDVector	SGDVector.subtract(SGDVector other)	Subtracts other from this vector, producing a new SGDVector.
SGDVector	SparseVector.subtract(SGDVector other)	Subtracts other from this vector, producing a new SGDVector.

Uses of SGDVector in org.tribuo.math.optimisers.util

Classes in org.tribuo.math.optimisers.util that implement SGDVector

Modifier and Type	Class	Description
class	ShrinkingVector	A subclass of DenseVector which shrinks the value every time a new value is added.

Methods in org.tribuo.math.optimisers.util with parameters of type SGDVector

Modifier and Type	Method	Description
double	ShrinkingVector.dot(SGDVector other)
DenseVector	ShrinkingMatrix.leftMultiply(SGDVector input)

Uses of SGDVector in org.tribuo.multilabel.sgd

Subinterfaces with type arguments of type SGDVector in org.tribuo.multilabel.sgd

Modifier and Type	Interface	Description
interface	MultiLabelObjective	An interface for multi-label prediction objectives.

Uses of SGDVector in org.tribuo.multilabel.sgd.linear

Subclasses with type arguments of type SGDVector in org.tribuo.multilabel.sgd.linear

Modifier and Type	Class	Description
class	LinearSGDTrainer	A trainer for a multi-label linear model which uses SGD.

Methods in org.tribuo.multilabel.sgd.linear that return types with arguments of type SGDVector

Modifier and Type	Method	Description
protected SGDObjective<SGDVector>	LinearSGDTrainer.getObjective()

Uses of SGDVector in org.tribuo.multilabel.sgd.objectives

Methods in org.tribuo.multilabel.sgd.objectives that return types with arguments of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	BinaryCrossEntropy.lossAndGradient(SGDVector truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.lossAndGradient(SGDVector truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.

Methods in org.tribuo.multilabel.sgd.objectives with parameters of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	BinaryCrossEntropy.lossAndGradient(SGDVector truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Hinge.lossAndGradient(SGDVector truth, SGDVector prediction)	Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.

Uses of SGDVector in org.tribuo.regression.sgd

Methods in org.tribuo.regression.sgd that return types with arguments of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	RegressionObjective.loss(DenseVector truth, SGDVector prediction)	Deprecated. In 4.1 to move to the new name, lossAndGradient.
default com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	RegressionObjective.lossAndGradient(DenseVector truth, SGDVector prediction)

Methods in org.tribuo.regression.sgd with parameters of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	RegressionObjective.loss(DenseVector truth, SGDVector prediction)	Deprecated. In 4.1 to move to the new name, lossAndGradient.
default com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	RegressionObjective.lossAndGradient(DenseVector truth, SGDVector prediction)

Uses of SGDVector in org.tribuo.regression.sgd.objectives

Methods in org.tribuo.regression.sgd.objectives that return types with arguments of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	AbsoluteLoss.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Huber.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SquaredLoss.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	AbsoluteLoss.lossAndGradient(DenseVector truth, SGDVector prediction)
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Huber.lossAndGradient(DenseVector truth, SGDVector prediction)
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SquaredLoss.lossAndGradient(DenseVector truth, SGDVector prediction)

Methods in org.tribuo.regression.sgd.objectives with parameters of type SGDVector

Modifier and Type	Method	Description
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	AbsoluteLoss.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Huber.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SquaredLoss.loss(DenseVector truth, SGDVector prediction)	Deprecated.
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	AbsoluteLoss.lossAndGradient(DenseVector truth, SGDVector prediction)
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	Huber.lossAndGradient(DenseVector truth, SGDVector prediction)
com.oracle.labs.mlrg.olcut.util.Pair<Double, SGDVector>	SquaredLoss.lossAndGradient(DenseVector truth, SGDVector prediction)