hex.deeplearning

Class Neurons.Input

java.lang.Object

hex.deeplearning.Neurons

hex.deeplearning.Neurons.Input

Enclosing class:

Neurons

public static class Neurons.Input
extends Neurons

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.

Nested Class Summary

Nested classes/interfaces inherited from class hex.deeplearning.Neurons

Neurons.ExpRectifier, Neurons.ExpRectifierDropout, Neurons.Input, Neurons.Linear, Neurons.Maxout, Neurons.MaxoutDropout, Neurons.Output, Neurons.Rectifier, Neurons.RectifierDropout, Neurons.Softmax, Neurons.Tanh, Neurons.TanhDropout

Field Summary

Fields inherited from class hex.deeplearning.Neurons

_a, _avg_a, _b, _bEA, _dropout, _e, _index, _input, _origa, _previous, _w, _wEA, params, units

Method Summary

Modifier and Type	Method and Description
protected void	bprop(int n) Back propagation of error terms stored in _e (for non-final layers)
protected void	fprop(long seed, boolean training, int n) Forward propagation
void	setInput(long seed, double[] data, int mb) One of two methods to set layer input values.
void	setInput(long seed, int[] numIds, double[] nums, int numcat, int[] cats, int mb) The second method used to set input layer values.

Methods inherited from class hex.deeplearning.Neurons

autoEncoderGradient, bpropOutputLayer, init, momentum, momentum, rate, setOutputLayerGradient, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Method Detail

bprop

protected void bprop(int n)

Description copied from class: Neurons

Back propagation of error terms stored in _e (for non-final layers)

Specified by:

bprop in class Neurons

fprop

protected void fprop(long seed,
                     boolean training,
                     int n)

Description copied from class: Neurons

Forward propagation

Specified by:

fprop in class Neurons

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

setInput

public void setInput(long seed,
                     double[] data,
                     int mb)

One of two methods to set layer input values. This one is for raw double data, e.g. for scoring

Parameters:

seed - For seeding the RNG inside (for input dropout)

data - Data (training columns and responses) to extract the training columns from to be mapped into the input neuron layer

mb - Mini-Batch index (which point inside this mini-batch)

setInput

public void setInput(long seed,
                     int[] numIds,
                     double[] nums,
                     int numcat,
                     int[] cats,
                     int mb)

The second method used to set input layer values. This one is used directly by FrameTask.processRow() and by the method above.

Parameters:

seed - For seeding the RNG inside (for input dropout)

nums - Array containing numerical values, can be NaN

numcat - Number of horizontalized categorical non-zero values (i.e., those not being the first factor of a class)

cats - Array of indices, the first numcat values are the input layer unit (==column) indices for the non-zero categorical values (This allows this array to be re-usable by the caller, without re-allocating each time)

mb - Mini-Batch index (which point inside this mini-batch)