|
| ||||
| |||||
| Package | Description |
|---|---|
| org.apache.ignite.ml |
Root ML package.
|
| org.apache.ignite.ml.composition |
Contains classes for ensemble of models implementation.
|
| org.apache.ignite.ml.composition.boosting |
Contains Gradient Boosting regression and classification abstract classes
allowing regressor type selecting in child classes.
|
| org.apache.ignite.ml.composition.boosting.convergence |
Package contains implementation of convergency checking algorithms for gradient boosting.
|
| org.apache.ignite.ml.composition.boosting.convergence.mean |
Contains implementation of convergence checking computer by mean of absolute value of errors in dataset.
|
| org.apache.ignite.ml.composition.boosting.convergence.median |
Contains implementation of convergence checking computer by median of medians of errors in dataset.
|
| org.apache.ignite.ml.composition.boosting.convergence.simple |
Contains implementation of Stub for convergence checking.
|
| org.apache.ignite.ml.composition.combinators.sequential |
Contains sequential combinators of trainers and models.
|
| org.apache.ignite.ml.composition.predictionsaggregator |
Contains classes for several predictions aggregation strategies
working with predictions vector from models ensemble.
|
| org.apache.ignite.ml.composition.stacking |
Contains classes used for training with stacking technique.
|
| org.apache.ignite.ml.dataset |
Base package for machine learning dataset classes.
|
| org.apache.ignite.ml.dataset.impl.cache.util |
Contains util classes used in cache based implementation of dataset.
|
| org.apache.ignite.ml.environment |
Package contains environment utils for ML algorithms.
|
| org.apache.ignite.ml.environment.logging |
Package contains several logging strategy realisations.
|
| org.apache.ignite.ml.environment.parallelism |
Package contains realisations of parallelism strategies for multi-thread algorithms.
|
| org.apache.ignite.ml.math.functions |
Contains serializable functions for distributed code algebra.
|
| org.apache.ignite.ml.math.primitives.matrix |
Contains matrix related classes.
|
| org.apache.ignite.ml.math.primitives.vector |
Contains vector related classes.
|
| org.apache.ignite.ml.nn |
Contains neural networks and related classes.
|
| org.apache.ignite.ml.optimization |
Contains implementations of optimization algorithms and related classes.
|
| org.apache.ignite.ml.optimization.updatecalculators |
Contains update calculators.
|
| org.apache.ignite.ml.preprocessing |
Base package for machine learning preprocessing classes.
|
| org.apache.ignite.ml.preprocessing.developer |
Contains Developer API preprocessors.
|
| org.apache.ignite.ml.trainers |
Contains model trainers.
|
| org.apache.ignite.ml.tree.randomforest.data |
Package contains helper data structures for random forest implementation.
|
| org.apache.ignite.ml.util.generators |
Contains utility classes for data streams generation.
|
| org.apache.ignite.ml.util.generators.primitives.scalar |
Contains generators of pseudo-random scalars in according to specific disctribution.
|
| org.apache.ignite.ml.util.generators.primitives.vector |
Contains generators of pseudo-random vectors in according to specific disctribution.
|
| org.apache.ignite.ml.util.generators.standard |
Contains classes for predefined data stream generators.
|
| Modifier and Type | Method and Description |
|---|---|
default <V1> IgniteModel<V1,V> |
IgniteModel.andBefore(IgniteFunction<V1,T> before)
Get a composition model of the form
x -> mdl(before(x)). |
default <V1> IgniteModel<T,V1> |
IgniteModel.andThen(IgniteFunction<V,V1> after)
Get a composition model of the form
x -> after(mdl(x)). |
| Modifier and Type | Method and Description |
|---|---|
static <L> DatasetMapping<L,L> |
DatasetMapping.mappingFeatures(IgniteFunction<Vector,Vector> mapper)
Dataset mapping which maps features, leaving labels unaffected.
|
| Modifier and Type | Field and Description |
|---|---|
protected IgniteFunction<Double,Double> |
GDBLearningStrategy.externalLbToInternalMapping
External label to internal mapping.
|
| Modifier and Type | Method and Description |
|---|---|
GDBLearningStrategy |
GDBLearningStrategy.withExternalLabelToInternal(IgniteFunction<Double,Double> externalLbToInternal)
Sets external to internal label representation mapping.
|
| Constructor and Description |
|---|
GDBModel(List<? extends IgniteModel<Vector,Double>> models,
WeightedPredictionsAggregator predictionsAggregator,
IgniteFunction<Double,Double> internalToExternalLblMapping)
Creates an instance of GDBModel.
|
| Modifier and Type | Method and Description |
|---|---|
abstract <K,V> ConvergenceChecker<K,V> |
ConvergenceCheckerFactory.create(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> vectorizer)
Create an instance of ConvergenceChecker.
|
| Constructor and Description |
|---|
ConvergenceChecker(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor,
double precision)
Constructs an instance of ConvergenceChecker.
|
| Modifier and Type | Method and Description |
|---|---|
<K,V> ConvergenceChecker<K,V> |
MeanAbsValueConvergenceCheckerFactory.create(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor)
Create an instance of ConvergenceChecker.
|
| Constructor and Description |
|---|
MeanAbsValueConvergenceChecker(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor,
double precision)
Creates an instance of MeanAbsValueConvergenceChecker.
|
| Modifier and Type | Method and Description |
|---|---|
<K,V> ConvergenceChecker<K,V> |
MedianOfMedianConvergenceCheckerFactory.create(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor)
Create an instance of ConvergenceChecker.
|
| Constructor and Description |
|---|
MedianOfMedianConvergenceChecker(long sampleSize,
IgniteFunction<Double,Double> lblMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor,
double precision)
Creates an instance of MedianOfMedianConvergenceChecker.
|
| Modifier and Type | Method and Description |
|---|---|
<K,V> ConvergenceChecker<K,V> |
ConvergenceCheckerStubFactory.create(long sampleSize,
IgniteFunction<Double,Double> externalLbToInternalMapping,
Loss loss,
DatasetBuilder<K,V> datasetBuilder,
Preprocessor<K,V> preprocessor)
Create an instance of ConvergenceChecker.
|
| Constructor and Description |
|---|
ConvergenceCheckerStub(long sampleSize,
IgniteFunction externalLbToInternalMapping,
Loss loss,
DatasetBuilder datasetBuilder,
Preprocessor<K,V> preprocessor,
double precision)
Creates an instance of ConvergenceCheckerStub.
|
| Modifier and Type | Field and Description |
|---|---|
protected IgniteBiFunction<Integer,? super IgniteModel<I,O1>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> |
TrainersSequentialComposition.datasetMapping
Dataset mapping.
|
| Modifier and Type | Method and Description |
|---|---|
static <I,O,L> TrainersSequentialComposition<I,O,O,L> |
TrainersSequentialComposition.ofSame(DatasetTrainer<? extends IgniteModel<I,O>,L> tr,
IgniteBiFunction<Integer,? super IgniteModel<I,O>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMapping,
IgniteBiPredicate<Integer,IgniteModel<I,O>> shouldStop,
IgniteFunction<O,I> out2In)
Construct sequential composition of same trainers.
|
static <I,O> ModelsSequentialComposition<I,O,O> |
ModelsSequentialComposition.ofSame(List<? extends IgniteModel<I,O>> lst,
IgniteFunction<O,I> output2Input)
Get sequential composition of submodels with same type.
|
| Modifier and Type | Method and Description |
|---|---|
static <I,O,L> TrainersSequentialComposition<I,O,O,L> |
TrainersSequentialComposition.ofSame(DatasetTrainer<? extends IgniteModel<I,O>,L> tr,
IgniteBiFunction<Integer,? super IgniteModel<I,O>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMapping,
IgniteBiPredicate<Integer,IgniteModel<I,O>> shouldStop,
IgniteFunction<O,I> out2In)
Construct sequential composition of same trainers.
|
| Constructor and Description |
|---|
TrainersSequentialComposition(DatasetTrainer<? extends IgniteModel<I,O1>,L> tr1,
DatasetTrainer<? extends IgniteModel<O1,O2>,L> tr2,
IgniteFunction<? super IgniteModel<I,O1>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMapping)
Construct sequential composition of given two trainers.
|
| Constructor and Description |
|---|
TrainersSequentialComposition(DatasetTrainer<? extends IgniteModel<I,O1>,L> tr1,
DatasetTrainer<? extends IgniteModel<O1,O2>,L> tr2,
IgniteBiFunction<Integer,? super IgniteModel<I,O1>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMapping)
Create sequential composition of two trainers.
|
TrainersSequentialComposition(DatasetTrainer<? extends IgniteModel<I,O1>,L> tr1,
DatasetTrainer<? extends IgniteModel<O1,O2>,L> tr2,
IgniteFunction<? super IgniteModel<I,O1>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMapping)
Construct sequential composition of given two trainers.
|
| Modifier and Type | Interface and Description |
|---|---|
interface |
PredictionsAggregator
Predictions aggregator interface.
|
| Modifier and Type | Class and Description |
|---|---|
class |
MeanValuePredictionsAggregator
Predictions aggregator returning the mean value of predictions.
|
class |
OnMajorityPredictionsAggregator
Predictions aggregator returning the most frequently prediction.
|
class |
WeightedPredictionsAggregator
Predictions aggregator returning weighted plus of predictions.
|
| Modifier and Type | Method and Description |
|---|---|
<L1> StackedVectorDatasetTrainer<O,AM,L1> |
StackedVectorDatasetTrainer.withConvertedLabels(IgniteFunction<L1,L> new2Old)
Creates
DatasetTrainer with same training logic, but able to accept labels of given new type of labels. |
<L1> SimpleStackedDatasetTrainer<I,O,AM,L1> |
SimpleStackedDatasetTrainer.withConvertedLabels(IgniteFunction<L1,L> new2Old)
Creates
DatasetTrainer with same training logic, but able to accept labels of given new type of labels. |
SimpleStackedDatasetTrainer<I,O,AM,L> |
SimpleStackedDatasetTrainer.withOriginalFeaturesKept(IgniteFunction<I,I> submodelInput2AggregatingInputConverter)
Keep original features during training and propagate submodels input to aggregator during inference
using given function.
|
StackedDatasetTrainer<IS,IA,O,AM,L> |
StackedDatasetTrainer.withOriginalFeaturesKept(IgniteFunction<IS,IA> submodelInput2AggregatingInputConverter)
Keep original features during training and propagate submodels input to aggregator during inference
using given function.
|
StackedVectorDatasetTrainer<O,AM,L> |
StackedVectorDatasetTrainer.withOriginalFeaturesKept(IgniteFunction<Vector,Vector> submodelInput2AggregatingInputConverter)
Keep original features during training and propagate submodels input to aggregator during inference
using given function.
|
StackedDatasetTrainer<IS,IA,O,AM,L> |
StackedDatasetTrainer.withSubmodelOutput2VectorConverter(IgniteFunction<IA,Vector> submodelOutput2VectorConverter)
Set function used for conversion of submodel output to
Vector. |
StackedVectorDatasetTrainer<O,AM,L> |
StackedVectorDatasetTrainer.withSubmodelOutput2VectorConverter(IgniteFunction<Vector,Vector> submodelOutput2VectorConverter)
Set function used for conversion of submodel output to
Vector. |
StackedDatasetTrainer<IS,IA,O,AM,L> |
StackedDatasetTrainer.withVector2SubmodelInputConverter(IgniteFunction<Vector,IS> vector2SubmodelInputConverter)
Set function used for conversion of
Vector to submodel input. |
StackedVectorDatasetTrainer<O,AM,L> |
StackedVectorDatasetTrainer.withVector2SubmodelInputConverter(IgniteFunction<Vector,Vector> vector2SubmodelInputConverter)
Set function used for conversion of
Vector to submodel input. |
| Constructor and Description |
|---|
SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer,
IgniteBinaryOperator<I> aggregatingInputMerger,
IgniteFunction<I,I> submodelInput2AggregatingInputConverter,
IgniteFunction<Vector,I> vector2SubmodelInputConverter,
IgniteFunction<I,Vector> submodelOutput2VectorConverter)
Construct instance of this class.
|
SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer,
IgniteBinaryOperator<I> aggregatingInputMerger,
IgniteFunction<I,I> submodelInput2AggregatingInputConverter,
IgniteFunction<Vector,I> vector2SubmodelInputConverter,
IgniteFunction<I,Vector> submodelOutput2VectorConverter)
Construct instance of this class.
|
SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer,
IgniteBinaryOperator<I> aggregatingInputMerger,
IgniteFunction<I,I> submodelInput2AggregatingInputConverter,
IgniteFunction<Vector,I> vector2SubmodelInputConverter,
IgniteFunction<I,Vector> submodelOutput2VectorConverter)
Construct instance of this class.
|
StackedDatasetTrainer(DatasetTrainer<AM,L> aggregatorTrainer,
IgniteBinaryOperator<IA> aggregatingInputMerger,
IgniteFunction<IS,IA> submodelInput2AggregatingInputConverter)
Constructs instance of this class.
|
StackedDatasetTrainer(DatasetTrainer<AM,L> aggregatorTrainer,
IgniteBinaryOperator<IA> aggregatingInputMerger,
IgniteFunction<IS,IA> submodelInput2AggregatingInputConverter,
List<DatasetTrainer<IgniteModel<IS,IA>,L>> submodelsTrainers,
IgniteFunction<Vector,IS> vector2SubmodelInputConverter,
IgniteFunction<IA,Vector> submodelOutput2VectorConverter)
Create instance of this class.
|
StackedDatasetTrainer(DatasetTrainer<AM,L> aggregatorTrainer,
IgniteBinaryOperator<IA> aggregatingInputMerger,
IgniteFunction<IS,IA> submodelInput2AggregatingInputConverter,
List<DatasetTrainer<IgniteModel<IS,IA>,L>> submodelsTrainers,
IgniteFunction<Vector,IS> vector2SubmodelInputConverter,
IgniteFunction<IA,Vector> submodelOutput2VectorConverter)
Create instance of this class.
|
StackedDatasetTrainer(DatasetTrainer<AM,L> aggregatorTrainer,
IgniteBinaryOperator<IA> aggregatingInputMerger,
IgniteFunction<IS,IA> submodelInput2AggregatingInputConverter,
List<DatasetTrainer<IgniteModel<IS,IA>,L>> submodelsTrainers,
IgniteFunction<Vector,IS> vector2SubmodelInputConverter,
IgniteFunction<IA,Vector> submodelOutput2VectorConverter)
Create instance of this class.
|
| Modifier and Type | Method and Description |
|---|---|
default <C2 extends Serializable> |
PartitionContextBuilder.andThen(IgniteFunction<C,C2> fun)
Makes a composed partition
context builder that first builds a context and then applies the
specified function on the result. |
default <R> R |
Dataset.compute(IgniteFunction<D,R> map,
IgniteBinaryOperator<R> reduce)
Applies the specified
map function to every partition data in the dataset and then reduces
map results to final result by using the reduce function. |
default <R> R |
Dataset.compute(IgniteFunction<D,R> map,
IgniteBinaryOperator<R> reduce,
R identity)
Applies the specified
map function to every partition data in the dataset and then reduces
map results to final result by using the reduce function. |
default <I extends Dataset<C,D>> |
Dataset.wrap(IgniteFunction<Dataset<C,D>,I> wrapper)
Wraps this dataset into the specified wrapper to introduce new functionality based on
compute and
computeWithCtx methods. |
| Modifier and Type | Method and Description |
|---|---|
static <R> Collection<R> |
ComputeUtils.affinityCallWithRetries(Ignite ignite,
Collection<String> cacheNames,
IgniteFunction<Integer,R> fun,
int retries,
DeployingContext deployingContext)
Calls the specified
fun function on all partitions so that is't guaranteed that partitions with the same
index of all specified caches will be placed on the same node and will not be moved before computation is
finished. |
static <R> Collection<R> |
ComputeUtils.affinityCallWithRetries(Ignite ignite,
Collection<String> cacheNames,
IgniteFunction<Integer,R> fun,
int retries,
int interval,
DeployingContext deployingContext)
Calls the specified
fun function on all partitions so that is't guaranteed that partitions with the same
index of all specified caches will be placed on the same node and will not be moved before computation is
finished. |
| Modifier and Type | Method and Description |
|---|---|
DefaultLearningEnvironmentBuilder |
DefaultLearningEnvironmentBuilder.withLoggingFactoryDependency(IgniteFunction<Integer,MLLogger.Factory> loggingFactory)
Specify dependency (partition -> logging factory).
|
LearningEnvironmentBuilder |
LearningEnvironmentBuilder.withLoggingFactoryDependency(IgniteFunction<Integer,MLLogger.Factory> loggingFactory)
Specify dependency (partition -> logging factory).
|
DefaultLearningEnvironmentBuilder |
DefaultLearningEnvironmentBuilder.withParallelismStrategyDependency(IgniteFunction<Integer,ParallelismStrategy> stgy)
Specifies dependency (partition -> Parallelism Strategy for LearningEnvironment).
|
LearningEnvironmentBuilder |
LearningEnvironmentBuilder.withParallelismStrategyDependency(IgniteFunction<Integer,ParallelismStrategy> stgy)
Specifies dependency (partition -> Parallelism Strategy for LearningEnvironment).
|
DefaultLearningEnvironmentBuilder |
DefaultLearningEnvironmentBuilder.withParallelismStrategyTypeDependency(IgniteFunction<Integer,ParallelismStrategy.Type> stgyType)
Specifies dependency (partition -> Parallelism Strategy Type for LearningEnvironment).
|
LearningEnvironmentBuilder |
LearningEnvironmentBuilder.withParallelismStrategyTypeDependency(IgniteFunction<Integer,ParallelismStrategy.Type> stgyType)
Specifies dependency (partition -> Parallelism Strategy Type for LearningEnvironment).
|
LearningEnvironmentBuilder |
DefaultLearningEnvironmentBuilder.withRandomDependency(IgniteFunction<Integer,Random> rngSupplier)
Specify dependency (partition -> random numbers generator).
|
LearningEnvironmentBuilder |
LearningEnvironmentBuilder.withRandomDependency(IgniteFunction<Integer,Random> rngSupplier)
Specify dependency (partition -> random numbers generator).
|
LearningEnvironmentBuilder |
DefaultLearningEnvironmentBuilder.withRNGSeedDependency(IgniteFunction<Integer,Long> seed)
Specify dependency (partition -> seed for random number generator).
|
LearningEnvironmentBuilder |
LearningEnvironmentBuilder.withRNGSeedDependency(IgniteFunction<Integer,Long> seed)
Specify dependency (partition -> seed for random number generator).
|
| Modifier and Type | Field and Description |
|---|---|
static IgniteFunction<Integer,MLLogger.Factory> |
ConsoleLogger.Factory.HIGH
High.
|
static IgniteFunction<Integer,MLLogger.Factory> |
ConsoleLogger.Factory.LOW
Low.
|
static IgniteFunction<Integer,MLLogger.Factory> |
ConsoleLogger.Factory.OFF
Offset.
|
| Modifier and Type | Field and Description |
|---|---|
static IgniteFunction<Integer,ParallelismStrategy.Type> |
ParallelismStrategy.NO_PARALLELISM
No parallelism.
|
static IgniteFunction<Integer,ParallelismStrategy.Type> |
ParallelismStrategy.ON_DEFAULT_POOL
On default pool.
|
| Modifier and Type | Interface and Description |
|---|---|
interface |
IgniteCurriedBiFunction<A,B,T>
Serializable binary function.
|
interface |
IgniteCurriedTriFunction<A,B,C,D>
Serializable curried tri-function.
|
interface |
IgniteDifferentiableVectorToDoubleFunction
Interface for differentiable functions from vector to double.
|
| Modifier and Type | Method and Description |
|---|---|
default <V> IgniteFunction<T,V> |
IgniteFunction.andThen(IgniteFunction<? super R,? extends V> after)
Compose this function and given function.
|
static <T,R> IgniteFunction<T,R> |
IgniteFunction.constant(R r)
IgniteFunction returning specified constant. |
static <T> IgniteFunction<T,T> |
IgniteFunction.identity()
Identity function.
|
static <A,B> IgniteFunction<A,IgniteSupplier<B>> |
Functions.outputSupplier(IgniteFunction<A,B> f)
Transform function of form a -> b into a -> (() -> b).
|
| Modifier and Type | Method and Description |
|---|---|
default <V> IgniteBiFunction<T,U,V> |
IgniteBiFunction.andThen(IgniteFunction<? super R,? extends V> after) |
default <V> IgniteFunction<T,V> |
IgniteFunction.andThen(IgniteFunction<? super R,? extends V> after)
Compose this function and given function.
|
static <A,B extends Comparable<B>> |
Functions.argmin(List<A> args,
IgniteFunction<A,B> f) |
static <A,B> IgniteFunction<A,IgniteSupplier<B>> |
Functions.outputSupplier(IgniteFunction<A,B> f)
Transform function of form a -> b into a -> (() -> b).
|
| Modifier and Type | Method and Description |
|---|---|
Vector |
AbstractMatrix.foldColumns(IgniteFunction<Vector,Double> fun)
Collects the results of applying a given function to all columns in this matrix.
|
Vector |
Matrix.foldColumns(IgniteFunction<Vector,Double> fun)
Collects the results of applying a given function to all columns in this matrix.
|
Vector |
AbstractMatrix.foldRows(IgniteFunction<Vector,Double> fun)
Collects the results of applying a given function to all rows in this matrix.
|
Vector |
Matrix.foldRows(IgniteFunction<Vector,Double> fun)
Collects the results of applying a given function to all rows in this matrix.
|
| Modifier and Type | Method and Description |
|---|---|
static IgniteFunction<Vector,Vector> |
VectorUtils.getProjector(int[] mapping)
Get projector from index mapping.
|
| Modifier and Type | Method and Description |
|---|---|
IgniteFunction<List<U>,U> |
UpdatesStrategy.allUpdatesReducer()
Get function used to reduce updates from different trainings
(for example, averaging of gradients of all parallel trainings).
|
IgniteFunction<Dataset<EmptyContext,SimpleLabeledDatasetData>,MLPArchitecture> |
MLPTrainer.getArchSupplier()
Get the multilayer perceptron architecture supplier that defines layers and activators.
|
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
MLPTrainer.getLoss()
Get the loss function to be minimized during the training.
|
IgniteFunction<List<U>,U> |
UpdatesStrategy.locStepUpdatesReducer()
Get function used to reduce updates in one training
(for example, sum all sequential gradient updates to get one gradient update).
|
| Modifier and Type | Method and Description |
|---|---|
Vector |
MultilayerPerceptron.differentiateByParameters(IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss,
Matrix inputsBatch,
Matrix truthBatch)
Compose function in the following way: feed output of this model as input to second argument to loss function.
|
MLPTrainer<P> |
MLPTrainer.withArchSupplier(IgniteFunction<Dataset<EmptyContext,SimpleLabeledDatasetData>,MLPArchitecture> archSupplier)
Set up the multilayer perceptron architecture supplier that defines layers and activators.
|
MLPTrainer<P> |
MLPTrainer.withLoss(IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss)
Set up the loss function to be minimized during the training.
|
| Constructor and Description |
|---|
MLPTrainer(IgniteFunction<Dataset<EmptyContext,SimpleLabeledDatasetData>,MLPArchitecture> archSupplier,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss,
UpdatesStrategy<? super MultilayerPerceptron,P> updatesStgy,
int maxIterations,
int batchSize,
int locIterations,
long seed)
Constructs a new instance of multilayer perceptron trainer.
|
MLPTrainer(IgniteFunction<Dataset<EmptyContext,SimpleLabeledDatasetData>,MLPArchitecture> archSupplier,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss,
UpdatesStrategy<? super MultilayerPerceptron,P> updatesStgy,
int maxIterations,
int batchSize,
int locIterations,
long seed)
Constructs a new instance of multilayer perceptron trainer.
|
MLPTrainer(MLPArchitecture arch,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss,
UpdatesStrategy<? super MultilayerPerceptron,P> updatesStgy,
int maxIterations,
int batchSize,
int locIterations,
long seed)
Constructs a new instance of multilayer perceptron trainer.
|
UpdatesStrategy(ParameterUpdateCalculator<M,U> updatesCalculator,
IgniteFunction<List<U>,U> locStepUpdatesReducer,
IgniteFunction<List<U>,U> allUpdatesReducer)
Construct instance of this class with given parameters.
|
UpdatesStrategy(ParameterUpdateCalculator<M,U> updatesCalculator,
IgniteFunction<List<U>,U> locStepUpdatesReducer,
IgniteFunction<List<U>,U> allUpdatesReducer)
Construct instance of this class with given parameters.
|
| Modifier and Type | Field and Description |
|---|---|
static IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
LossFunctions.HINGE
Hinge loss function.
|
static IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
LossFunctions.L1
L1 loss function.
|
static IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
LossFunctions.L2
L2 loss function.
|
static IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
LossFunctions.LOG
Log loss function.
|
static IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
LossFunctions.MSE
Mean squared error loss function.
|
| Modifier and Type | Method and Description |
|---|---|
Vector |
SmoothParametrized.differentiateByParameters(IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss,
Matrix inputsBatch,
Matrix truthBatch)
Compose function in the following way: feed output of this model as input to second argument to loss function.
|
| Modifier and Type | Field and Description |
|---|---|
static IgniteFunction<List<RPropParameterUpdate>,RPropParameterUpdate> |
RPropParameterUpdate.AVG
Averages updates returned by different trainings.
|
static IgniteFunction<List<SimpleGDParameterUpdate>,SimpleGDParameterUpdate> |
SimpleGDParameterUpdate.AVG
Method used to sum updates inside of one of parallel trainings.
|
protected IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
RPropUpdateCalculator.loss
Loss function.
|
protected IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> |
SimpleGDUpdateCalculator.loss
Loss function.
|
static IgniteFunction<List<RPropParameterUpdate>,RPropParameterUpdate> |
RPropParameterUpdate.SUM
Sums updates returned by different trainings.
|
static IgniteFunction<List<RPropParameterUpdate>,RPropParameterUpdate> |
RPropParameterUpdate.SUM_LOCAL
Sums updates during one training.
|
static IgniteFunction<List<SimpleGDParameterUpdate>,SimpleGDParameterUpdate> |
SimpleGDParameterUpdate.SUM_LOCAL
Method used to get total update of all parallel trainings.
|
| Modifier and Type | Method and Description |
|---|---|
P |
ParameterUpdateCalculator.init(M mdl,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss)
Initializes the update calculator.
|
NesterovParameterUpdate |
NesterovUpdateCalculator.init(M mdl,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss)
Initializes the update calculator.
|
RPropParameterUpdate |
RPropUpdateCalculator.init(SmoothParametrized mdl,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss)
Initializes the update calculator.
|
SimpleGDParameterUpdate |
SimpleGDUpdateCalculator.init(SmoothParametrized mdl,
IgniteFunction<Vector,IgniteDifferentiableVectorToDoubleFunction> loss)
Initializes the update calculator.
|
| Modifier and Type | Method and Description |
|---|---|
default <L1,L2> Preprocessor<K,V> |
Preprocessor.map(IgniteFunction<LabeledVector<L1>,LabeledVector<L2>> func)
Map vectorizer answer.
|
| Constructor and Description |
|---|
MappedPreprocessor(Preprocessor<K,V> original,
IgniteFunction<LabeledVector<L0>,LabeledVector<L1>> andThen)
Creates an instance of MappedPreprocessor.
|
PatchedPreprocessor(IgniteFunction<LabeledVector<L1>,LabeledVector<L2>> lbPatcher,
Preprocessor<K,V> basePreprocessor)
Constructs a new instance of Binarization preprocessor.
|
| Modifier and Type | Method and Description |
|---|---|
AdaptableDatasetTrainer<I,O,IW,OW,M,L> |
AdaptableDatasetTrainer.afterFeatureExtractor(IgniteFunction<Vector,Vector> after)
Specify function which will be applied after feature extractor.
|
AdaptableDatasetTrainer<I,O,IW,OW,M,L> |
AdaptableDatasetTrainer.afterLabelExtractor(IgniteFunction<L,L> after)
Specify function which will be applied after label extractor.
|
<O1> AdaptableDatasetTrainer<I,O1,IW,OW,M,L> |
AdaptableDatasetTrainer.afterTrainedModel(IgniteFunction<O,O1> after)
Let this trainer produce model
mdl. |
<I1> AdaptableDatasetModel<I1,O,IW,OW,M> |
AdaptableDatasetModel.andBefore(IgniteFunction<I1,I> before)
Create new
AdaptableDatasetModel which is a composition of the form thisMdl . before. |
<O1,M1 extends IgniteModel<O,O1>> |
AdaptableDatasetTrainer.andThen(DatasetTrainer<M1,L> tr,
IgniteFunction<AdaptableDatasetModel<I,O,IW,OW,M>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMappingProducer)
Create a
TrainersSequentialComposition of whis trainer and specified trainer. |
<I1> AdaptableDatasetTrainer<I1,O,IW,OW,M,L> |
AdaptableDatasetTrainer.beforeTrainedModel(IgniteFunction<I1,I> before)
Let this trainer produce model
mdl. |
<L1> DatasetTrainer<M,L1> |
DatasetTrainer.withConvertedLabels(IgniteFunction<L1,L> new2Old)
Creates
DatasetTrainer with same training logic, but able to accept labels of given new type of labels. |
| Modifier and Type | Method and Description |
|---|---|
<O1,M1 extends IgniteModel<O,O1>> |
AdaptableDatasetTrainer.andThen(DatasetTrainer<M1,L> tr,
IgniteFunction<AdaptableDatasetModel<I,O,IW,OW,M>,IgniteFunction<LabeledVector<L>,LabeledVector<L>>> datasetMappingProducer)
Create a
TrainersSequentialComposition of whis trainer and specified trainer. |
| Constructor and Description |
|---|
AdaptableDatasetModel(IgniteFunction<I,IW> before,
M mdl,
IgniteFunction<OW,O> after)
Construct instance of this class.
|
AdaptableDatasetModel(IgniteFunction<I,IW> before,
M mdl,
IgniteFunction<OW,O> after)
Construct instance of this class.
|
| Modifier and Type | Field and Description |
|---|---|
static IgniteFunction<List<FeatureMeta>,Integer> |
FeaturesCountSelectionStrategies.ALL |
static IgniteFunction<List<FeatureMeta>,Integer> |
FeaturesCountSelectionStrategies.LOG2 |
static IgniteFunction<List<FeatureMeta>,Integer> |
FeaturesCountSelectionStrategies.ONE_THIRD |
static IgniteFunction<List<FeatureMeta>,Integer> |
FeaturesCountSelectionStrategies.SQRT |
| Modifier and Type | Method and Description |
|---|---|
default Stream<LabeledVector<Double>> |
DataStreamGenerator.labeled(IgniteFunction<Vector,Double> classifier) |
default DataStreamGenerator |
DataStreamGenerator.mapVectors(IgniteFunction<Vector,Vector> f)
Apply user defined mapper to vectors stream without labels hiding.
|
| Modifier and Type | Method and Description |
|---|---|
default IgniteFunction<Double,Double> |
RandomProducer.noizify(IgniteFunction<Double,Double> f)
Adds value generated by random producer to function value.
|
| Modifier and Type | Method and Description |
|---|---|
default IgniteFunction<Double,Double> |
RandomProducer.noizify(IgniteFunction<Double,Double> f)
Adds value generated by random producer to function value.
|
| Modifier and Type | Method and Description |
|---|---|
VectorGeneratorsFamily.Builder |
VectorGeneratorsFamily.Builder.map(IgniteFunction<VectorGenerator,VectorGenerator> mapper)
Adds map function for all generators in family.
|
default VectorGenerator |
VectorGenerator.map(IgniteFunction<Vector,Vector> mapper)
Maps values of vector generator using mapper.
|
| Constructor and Description |
|---|
ParametricVectorGenerator(RandomProducer paramGenerator,
IgniteFunction<Double,Double>... perDimensionGenerators)
Create an intance of ParametricVectorGenerator.
|
| Modifier and Type | Method and Description |
|---|---|
static RegressionDataStream |
RegressionDataStream.twoDimensional(IgniteFunction<Double,Double> function,
double minXVal,
double maxXVal)
Creates two dimensional regression data stream.
|
static RegressionDataStream |
RegressionDataStream.twoDimensional(IgniteFunction<Double,Double> function,
double minXVal,
double maxXVal,
long seed)
Creates two dimensional regression data stream.
|
| Constructor and Description |
|---|
RegressionDataStream(int vectorSize,
IgniteFunction<Vector,Double> function,
double minXVal,
double maxXVal)
Creates an instance of RegressionDataStream.
|
GridGain In-Memory Computing Platform : ver. 8.9.15 Release Date : December 3 2024