public class DeepLearningParametersV3 extends ModelParametersSchemaV3
| Modifier and Type | Field and Description |
|---|---|
DeepLearningActivation |
activation
Activation function.
|
boolean |
adaptiveRate
Adaptive learning rate.
|
boolean |
autoencoder
Auto-Encoder.
|
double |
averageActivation
Average activation for sparse auto-encoder.
|
boolean |
balanceClasses
Balance training data class counts via over/under-sampling (for imbalanced data).
|
double |
classificationStop
Stopping criterion for classification error fraction on training data (-1 to disable).
|
float[] |
classSamplingFactors
Desired over/under-sampling ratios per class (in lexicographic order).
|
boolean |
colMajor
#DEPRECATED Use a column major weight matrix for input layer.
|
boolean |
diagnostics
Enable diagnostics for hidden layers.
|
boolean |
elasticAveraging
Elastic averaging between compute nodes can improve distributed model convergence.
|
double |
elasticAveragingMovingRate
Elastic averaging moving rate (only if elastic averaging is enabled).
|
double |
elasticAveragingRegularization
Elastic averaging regularization strength (only if elastic averaging is enabled).
|
double |
epochs
How many times the dataset should be iterated (streamed), can be fractional.
|
double |
epsilon
Adaptive learning rate smoothing factor (to avoid divisions by zero and allow progress).
|
boolean |
exportWeightsAndBiases
Whether to export Neural Network weights and biases to H2O Frames.
|
boolean |
fastMode
Enable fast mode (minor approximation in back-propagation).
|
boolean |
forceLoadBalance
Force extra load balancing to increase training speed for small datasets (to keep all cores busy).
|
int[] |
hidden
Hidden layer sizes (e.g.
|
double[] |
hiddenDropoutRatios
Hidden layer dropout ratios (can improve generalization), specify one value per hidden layer, defaults to 0.5.
|
FrameKeyV3[] |
initialBiases
A list of H2OFrame ids to initialize the bias vectors of this model with.
|
DeepLearningInitialWeightDistribution |
initialWeightDistribution
Initial weight distribution.
|
FrameKeyV3[] |
initialWeights
A list of H2OFrame ids to initialize the weight matrices of this model with.
|
double |
initialWeightScale
Uniform: -value...value, Normal: stddev.
|
double |
inputDropoutRatio
Input layer dropout ratio (can improve generalization, try 0.1 or 0.2).
|
double |
l1
L1 regularization (can add stability and improve generalization, causes many weights to become 0).
|
double |
l2
L2 regularization (can add stability and improve generalization, causes many weights to be small.
|
DeepLearningLoss |
loss
Loss function.
|
float |
maxAfterBalanceSize
Maximum relative size of the training data after balancing class counts (can be less than 1.0).
|
int |
maxCategoricalFeatures
Max.
|
int |
maxConfusionMatrixSize
[Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs.
|
int |
maxHitRatioK
Max.
|
float |
maxW2
Constraint for squared sum of incoming weights per unit (e.g.
|
int |
miniBatchSize
Mini-batch size (smaller leads to better fit, larger can speed up and generalize better).
|
DeepLearningMissingValuesHandling |
missingValuesHandling
Handling of missing values.
|
double |
momentumRamp
Number of training samples for which momentum increases.
|
double |
momentumStable
Final momentum after the ramp is over (try 0.99).
|
double |
momentumStart
Initial momentum at the beginning of training (try 0.5).
|
boolean |
nesterovAcceleratedGradient
Use Nesterov accelerated gradient (recommended).
|
boolean |
overwriteWithBestModel
If enabled, override the final model with the best model found during training.
|
ModelKeyV3 |
pretrainedAutoencoder
Pretrained autoencoder model to initialize this model with.
|
boolean |
quietMode
Enable quiet mode for less output to standard output.
|
double |
rate
Learning rate (higher => less stable, lower => slower convergence).
|
double |
rateAnnealing
Learning rate annealing: rate / (1 + rate_annealing * samples).
|
double |
rateDecay
Learning rate decay factor between layers (N-th layer: rate * rate_decay ^ (n - 1).
|
double |
regressionStop
Stopping criterion for regression error (MSE) on training data (-1 to disable).
|
boolean |
replicateTrainingData
Replicate the entire training dataset onto every node for faster training on small datasets.
|
boolean |
reproducible
Force reproducibility on small data (will be slow - only uses 1 thread).
|
double |
rho
Adaptive learning rate time decay factor (similarity to prior updates).
|
double |
scoreDutyCycle
Maximum duty cycle fraction for scoring (lower: more training, higher: more scoring).
|
double |
scoreInterval
Shortest time interval (in seconds) between model scoring.
|
long |
scoreTrainingSamples
Number of training set samples for scoring (0 for all).
|
long |
scoreValidationSamples
Number of validation set samples for scoring (0 for all).
|
DeepLearningClassSamplingMethod |
scoreValidationSampling
Method used to sample validation dataset for scoring.
|
long |
seed
Seed for random numbers (affects sampling) - Note: only reproducible when running single threaded.
|
boolean |
shuffleTrainingData
Enable shuffling of training data (recommended if training data is replicated and train_samples_per_iteration is
close to #nodes x #rows, of if using balance_classes).
|
boolean |
singleNodeMode
Run on a single node for fine-tuning of model parameters.
|
boolean |
sparse
Sparse data handling (more efficient for data with lots of 0 values).
|
double |
sparsityBeta
Sparsity regularization.
|
boolean |
standardize
If enabled, automatically standardize the data.
|
double |
targetRatioCommToComp
Target ratio of communication overhead to computation.
|
long |
trainSamplesPerIteration
Number of training samples (globally) per MapReduce iteration.
|
boolean |
useAllFactorLevels
Use all factor levels of categorical variables.
|
boolean |
variableImportances
Compute variable importances for input features (Gedeon method) - can be slow for large networks.
|
categoricalEncoding, checkpoint, distribution, foldAssignment, foldColumn, huberAlpha, ignoreConstCols, ignoredColumns, keepCrossValidationFoldAssignment, keepCrossValidationPredictions, maxRuntimeSecs, modelId, nfolds, offsetColumn, parallelizeCrossValidation, quantileAlpha, responseColumn, scoreEachIteration, stoppingMetric, stoppingRounds, stoppingTolerance, trainingFrame, tweediePower, validationFrame, weightsColumn| Constructor and Description |
|---|
DeepLearningParametersV3()
Public constructor
|
| Modifier and Type | Method and Description |
|---|---|
java.lang.String |
toString()
Return the contents of this object as a JSON String.
|
public boolean balanceClasses
public float[] classSamplingFactors
public float maxAfterBalanceSize
public int maxConfusionMatrixSize
public int maxHitRatioK
public DeepLearningActivation activation
public int[] hidden
public double epochs
public long trainSamplesPerIteration
public double targetRatioCommToComp
public long seed
public boolean adaptiveRate
public double rho
public double epsilon
public double rate
public double rateAnnealing
public double rateDecay
public double momentumStart
public double momentumRamp
public double momentumStable
public boolean nesterovAcceleratedGradient
public double inputDropoutRatio
public double[] hiddenDropoutRatios
public double l1
public double l2
public float maxW2
public DeepLearningInitialWeightDistribution initialWeightDistribution
public double initialWeightScale
public FrameKeyV3[] initialWeights
public FrameKeyV3[] initialBiases
public DeepLearningLoss loss
public double scoreInterval
public long scoreTrainingSamples
public long scoreValidationSamples
public double scoreDutyCycle
public double classificationStop
public double regressionStop
public boolean quietMode
public DeepLearningClassSamplingMethod scoreValidationSampling
public boolean overwriteWithBestModel
public boolean autoencoder
public boolean useAllFactorLevels
public boolean standardize
public boolean diagnostics
public boolean variableImportances
public boolean fastMode
public boolean forceLoadBalance
public boolean replicateTrainingData
public boolean singleNodeMode
public boolean shuffleTrainingData
public DeepLearningMissingValuesHandling missingValuesHandling
public boolean sparse
public boolean colMajor
public double averageActivation
public double sparsityBeta
public int maxCategoricalFeatures
public boolean reproducible
public boolean exportWeightsAndBiases
public int miniBatchSize
public boolean elasticAveraging
public double elasticAveragingMovingRate
public double elasticAveragingRegularization
public ModelKeyV3 pretrainedAutoencoder
public java.lang.String toString()
toString in class ModelParametersSchemaV3