neuroidnet.ntr
Class SRMNeuroid

java.lang.Object
  |
  +--neuroidnet.ntr.Neuroid
        |
        +--neuroidnet.ntr.SRMNeuroid

All Implemented Interfaces:

Expressive, Input, Serializable, Simulation

Direct Known Subclasses:

AdThNeuroid, ArtificialConcept, PeakerNeuroid, SensoryNeuroid

public class SRMNeuroid

extends Neuroid

First attempt to implement the central neuroidal entity with programmability capability as defined by Valiant. Also includes simple soma functionality of a spiking neuron model, specifically the spike response model as defined in Maass and Bishop 1999.

Created: Thu May 2 22:00:37 2002

Modified: $Date: 2003/03/09 19:26:17 $

Version:

$Revision: 1.6 $ for this file.

Author:

Cengiz Gunay

See Also:

Serialized Form

Nested Class Summary

protected class

SRMNeuroid.Mode

Nested classes inherited from class neuroidnet.ntr.Neuroid

Neuroid.SynapseActivityTask

Field Summary

(package private) static double	correctTimesRequired Learning parameter for the Winnow learning algorithm.
(package private) double	externalCurrent Constant external current for all neuroids in network (not used!)
(package private) double	refractoryTimeConstant
(package private) double	sumOfCurrentWeights Internal.

Fields inherited from class neuroidnet.ntr.Neuroid

area, concept, debug, id, mode, potential, profile, spikeTrain, synapses, timeLastFired, watch

Constructor Summary

SRMNeuroid(Area area)
Sets mode and calls super().

SRMNeuroid(Area area, double initialThreshold, double refractoryTimeConstant)
Sets initial parameters.

Method Summary

protected void	calculatePotential() Iterate through Synapses and add kernel values and decrease refractoriness at this time value.
static double	defaultPeriod() Helper method to get a period for initiating calculations in Network.
void	fire() Saves firing time to spikeTrain.
double	getRefractoryTimeConstant() Get the value of refractoryTimeConstant.
(package private) double	refractoriness(double time) Refractory kernel taken from Maass and Bishop 1999, Eq.
void	setRefractoryTimeConstant(double v) Set the value of refractoryTimeConstant.

Methods inherited from class neuroidnet.ntr.Neuroid

addSynapse, dumpData, getArea, getConcept, getId, getMode, getProfile, getProperties, getSpikeTrain, getStatus, getSynapses, getTimeLastFired, init, isDebug, isWatch, makeConcept, setArea, setDebug, setMode, setProfile, setSpikeTrain, setWatch, step, stop, toString

Methods inherited from class java.lang.Object

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

Field Detail

correctTimesRequired

static final double correctTimesRequired

Learning parameter for the Winnow learning algorithm.

See Also:

SRMNeuroid.SRMMode#updateWeights, Constant Field Values

refractoryTimeConstant

double refractoryTimeConstant

externalCurrent

double externalCurrent

Constant external current for all neuroids in network (not used!)

sumOfCurrentWeights

double sumOfCurrentWeights

Internal.

Constructor Detail

SRMNeuroid

public SRMNeuroid(Area area)

Sets mode and calls super().

SRMNeuroid

public SRMNeuroid(Area area,
                  double initialThreshold,
                  double refractoryTimeConstant)

Sets initial parameters. (Not used: Calculates sumOfWeights from period length given by the Area.)

Parameters:

area - an Area value

refractoryTimeConstant -

See Also:

SRMNeuroid.Mode.sumOfWeights, Area.period

Method Detail

getRefractoryTimeConstant

public double getRefractoryTimeConstant()

Get the value of refractoryTimeConstant.

Returns:

value of refractoryTimeConstant.

setRefractoryTimeConstant

public void setRefractoryTimeConstant(double v)

Set the value of refractoryTimeConstant.

Parameters:

v - Value to assign to refractoryTimeConstant.

fire

public void fire()

Saves firing time to spikeTrain. And removes old spikes unless being watched.

Specified by:

fire in interface Input

Overrides:

fire in class Neuroid

See Also:

Area.fireNeuroid(neuroidnet.ntr.Neuroid), Neuroid.watch

defaultPeriod

public static double defaultPeriod()

Helper method to get a period for initiating calculations in Network. Initial period is calculated using network parameter values taken from Hopfield & Herz 95.

Returns:

a double value

See Also:

Network

refractoriness

double refractoriness(double time)

Refractory kernel taken from Maass and Bishop 1999, Eq. 1.47 on p.31. Keeps equivalence to integrate and fire model.

Parameters:

time - a double value

Returns:

a double potential value

calculatePotential

protected void calculatePotential()

Iterate through Synapses and add kernel values and decrease refractoriness at this time value. TODO: add external current!

Specified by:

calculatePotential in class Neuroid

See Also:

fire()