org.simbrain.network.neuron_update_rules.interfaces

Interface DifferentiableUpdateRule

All Known Implementing Classes:

AbstractSigmoidalRule, ContinuousSigmoidalRule, LinearRule, SigmoidalRule

public interface DifferentiableUpdateRule

Indicates that an update rule is differentiable, and has a getDerivative function. Used by backprop.

Author:

jyoshimi

Method Summary

Modifier and Type	Method and Description
double	getDerivative(double val) The inverse of the activation function.

Method Detail

getDerivative

double getDerivative(double val)

The inverse of the activation function.

Parameters:

val - the value being sent through the neuron's derivative

Returns:

the derivative of the neuron's activation function with respect to val.