deepboof.backward

Interface DFunctionDropOut<T extends deepboof.Tensor<T>>

All Superinterfaces:

deepboof.DFunction<T>, deepboof.Function<T>

All Known Implementing Classes:

DFunctionDropOut_F64

public interface DFunctionDropOut<T extends deepboof.Tensor<T>>
extends deepboof.DFunction<T>

Drop out is a technique introduced by [1] for regularizing a network and helps prevents over fitting. It works by randomly selecting neurons and forces them to be off. The chance of a neuron being turned off is specified by the drop rate. It's behavior is different when in learning or evaluation mode. In learning mode it will decide if a neuron is dropped using a probability of drop_rate*100, drop_rate is 0 to 1.0, inclusive. In evaluation mode it scales each input by 1.0 - drop_rate.

[1] Srivastava et al. "Dropout: A Simple Way to Prevent Neural Networks from Overfitting"

Method Summary

Modifier and Type	Method and Description
double	getDropRate() Returns a number from 0 to 1 indicating the likelihood of a neuron being dropped.

Methods inherited from interface deepboof.DFunction

backwards, evaluating, isLearning, learning

Methods inherited from interface deepboof.Function

forward, getOutputShape, getParameters, getParameterShapes, getTensorType, initialize, setParameters

Method Detail

getDropRate

double getDropRate()

Returns a number from 0 to 1 indicating the likelihood of a neuron being dropped. 0 = 0% change and 1 = 100% chance

Returns:

drop rate