Neurons (h2o-algos version 3.12.0.1 API)

java.lang.Object
- hex.deeplearning.Neurons

Direct Known Subclasses:

Neurons.ExpRectifier, Neurons.Input, Neurons.Maxout, Neurons.Output, Neurons.Rectifier, Neurons.Tanh
```
public abstract class Neurons
extends java.lang.Object
```
This class implements the concept of a Neuron layer in a Neural Network During training, every MRTask F/J thread is expected to create these neurons for every map call (Cheap to make). These Neurons are NOT sent over the wire. The weights connecting the neurons are in a separate class (DeepLearningModel.DeepLearningModelInfo), and will be shared per node.

Nested Class Summary

Nested Classes
Modifier and Type	Class and Description
`static class`	`Neurons.ExpRectifier`
`static class`	`Neurons.ExpRectifierDropout` Exponential Rectifier with dropout
`static class`	`Neurons.Input` Input layer of the Neural Network This layer is different from other layers as it has no incoming weights, but instead gets its activation values from the training points.
`static class`	`Neurons.Linear` Output neurons for regression - Linear units
`static class`	`Neurons.Maxout` Maxout neurons (picks the max out of the k activation_j = sum(A_ij*x_i) + b_j) Requires k times the model parameters (weights/biases) as a "normal" neuron
`static class`	`Neurons.MaxoutDropout` Maxout neurons with dropout
`static class`	`Neurons.Output` Abstract class for Output neurons
`static class`	`Neurons.Rectifier` Rectifier linear unit (ReLU) neurons
`static class`	`Neurons.RectifierDropout` Rectifier linear unit (ReLU) neurons with dropout
`static class`	`Neurons.Softmax` Output neurons for classification - Softmax
`static class`	`Neurons.Tanh` Tanh neurons - most common, most stable
`static class`	`Neurons.TanhDropout` Tanh neurons with dropout

Field Summary

Fields
Modifier and Type	Field and Description
`Storage.DenseVector[]`	`_a`
`Storage.DenseVector`	`_avg_a`
`Storage.DenseVector`	`_b`
`Storage.DenseVector`	`_bEA`
`protected Dropout`	`_dropout` For Dropout training
`Storage.DenseVector[]`	`_e`
`protected int`	`_index`
`Neurons`	`_input`
`Storage.DenseVector[]`	`_origa` Layer state (one per neuron): activity, error
`Neurons`	`_previous` References for feed-forward connectivity
`Storage.DenseRowMatrix`	`_w`
`Storage.DenseRowMatrix`	`_wEA`
`protected DeepLearningModel.DeepLearningParameters`	`params` Parameters (deep-cloned() from the user input, can be modified here, e.g.
`protected int`	`units`

Method Summary

Methods
Modifier and Type	Method and Description
`protected double`	`autoEncoderGradient(int row, int mb)` Helper to compute the reconstruction error for auto-encoders (part of the gradient computation)
`protected abstract void`	`bprop(int n)` Back propagation of error terms stored in _e (for non-final layers)
`protected void`	`bpropOutputLayer(int n)` Back-propagate gradient in output layer
`protected abstract void`	`fprop(long seed, boolean training, int n)` Forward propagation
`void`	`init(Neurons[] neurons, int index, DeepLearningModel.DeepLearningParameters p, DeepLearningModelInfo minfo, boolean training)` Initialization of the parameters and connectivity of a Neuron layer
`protected float`	`momentum()`
`float`	`momentum(double n)` The momentum - real number in [0, 1) Can be a linear ramp from momentum_start to momentum_stable, over momentum_ramp training samples
`float`	`rate(double n)` The learning rate
`protected void`	`setOutputLayerGradient(double ignored, int mb, int n)` Accumulation of reconstruction errors for a generic Neurons class (This is only used for AutoEncoders)
`java.lang.String`	`toString()` Print the status of this neuron layer

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

- Field Detail
  - units
```
protected int units
```
  - params
```
protected transient DeepLearningModel.DeepLearningParameters params
```
    Parameters (deep-cloned() from the user input, can be modified here, e.g. learning rate decay)
  - _index
```
protected transient int _index
```
  - _origa
```
public transient Storage.DenseVector[] _origa
```
    Layer state (one per neuron): activity, error
  - _a
```
public transient Storage.DenseVector[] _a
```
  - _e
```
public transient Storage.DenseVector[] _e
```
  - _previous
```
public Neurons _previous
```
    References for feed-forward connectivity
  - _input
```
public Neurons _input
```
  - _w
```
public Storage.DenseRowMatrix _w
```
  - _wEA
```
public Storage.DenseRowMatrix _wEA
```
  - _b
```
public Storage.DenseVector _b
```
  - _bEA
```
public Storage.DenseVector _bEA
```
  - _dropout
```
protected Dropout _dropout
```
    For Dropout training
  - _avg_a
```
public Storage.DenseVector _avg_a
```
- Method Detail
  - toString
```
public java.lang.String toString()
```
    Print the status of this neuron layer
    
    Overrides:
    
    toString in class java.lang.Object
    
    Returns:
    populated String
  - init
```
public final void init(Neurons[] neurons,
        int index,
        DeepLearningModel.DeepLearningParameters p,
        DeepLearningModelInfo minfo,
        boolean training)
```
    Initialization of the parameters and connectivity of a Neuron layer
    
    Parameters:
    neurons - Array of all neuron layers, to establish feed-forward connectivity
    index - Which layer am I?
    p - User-given parameters (Job parental object hierarchy is not used)
    minfo - Model information (weights/biases and their momenta)
    training - Whether training is done or just testing (no need for dropout)
  - fprop
```
protected abstract void fprop(long seed,
         boolean training,
         int n)
```
    Forward propagation
    
    Parameters:
    seed - For seeding the RNG inside (for dropout)
    training - Whether training is done or just testing (no need for dropout)
    n - number of actually trained samples in this mini-batch
  - bprop
```
protected abstract void bprop(int n)
```
    Back propagation of error terms stored in _e (for non-final layers)
  - bpropOutputLayer
```
protected final void bpropOutputLayer(int n)
```
    Back-propagate gradient in output layer
  - setOutputLayerGradient
```
protected void setOutputLayerGradient(double ignored,
                          int mb,
                          int n)
```
    Accumulation of reconstruction errors for a generic Neurons class (This is only used for AutoEncoders)
  - autoEncoderGradient
```
protected double autoEncoderGradient(int row,
                         int mb)
```
    Helper to compute the reconstruction error for auto-encoders (part of the gradient computation)
    
    Parameters:
    row - neuron index
    mb - minibatch-internal index
    
    Returns:
    difference between the output (auto-encoder output layer activation) and the target (input layer activation)
  - rate
```
public float rate(double n)
```
    The learning rate
    
    Parameters:
    n - The number of training samples seen so far (for rate_annealing greater than 0)
    
    Returns:
    Learning rate
  - momentum
```
protected float momentum()
```
  - momentum
```
public final float momentum(double n)
```
    The momentum - real number in [0, 1) Can be a linear ramp from momentum_start to momentum_stable, over momentum_ramp training samples
    
    Parameters:
    n - The number of training samples seen so far
    
    Returns:
    momentum

Class Neurons

Nested Class Summary

Field Summary

Method Summary

Methods inherited from class java.lang.Object

Field Detail

units

params

_index

_origa

_a

_e

_previous

_input

_w

_wEA

_b

_bEA

_dropout

_avg_a

Method Detail

toString

init

fprop

bprop

bpropOutputLayer

setOutputLayerGradient

autoEncoderGradient

rate

momentum

momentum