---

Layer Class Reference

Layer groups a number of Neurons for convenience. Layer is not needed to operate the network. More...

#include <amygdala/layer.h>

List of all members.

Public Types
typedef vector< Neuron * >::iterator	iterator
typedef vector< Neuron * >::const_iterator	const_iterator
typedef vector< Neuron * >::reverse_iterator	reverse_iterator
typedef vector< Neuron * >::const_reverse_iterator	const_reverse_iterator
Public Methods
	Layer ()
	~Layer ()
void	AddNeuron (Neuron *nrn)
void	AddNeuronVector (vector< Neuron * > nrnVec)
void	LayerName (string name)
string	LayerName ()
void	SetLayerId (unsigned int id)
unsigned int	LayerId ()
void	SetLayerType (LayerType ltype)
LayerType	GetLayerType ()
void	SetPercentInhibitory (float percent)
bool	ConnectLayers (Layer *output, UniConnectType parms, float pctInhibitory=0.0)
bool	ConnectLayers (Layer *output, GaussConnectType parms, float pctInhibitory=0.0)
bool	SetLayerConstants (LayerConstants lconst)
LayerConstants	GetLayerConstants ()
void	SetSynapticDelay (AmTimeInt delay)
AmTimeInt	GetSynapticDelay () const
unsigned int	size ()
iterator	begin ()
iterator	end ()
const_iterator	begin () const
const_iterator	end () const
reverse_iterator	rbegin ()
reverse_iterator	rend ()
const_reverse_iterator	rbegin () const
const_reverse_iterator	rend () const
Neuron *	operator[] (unsigned int &index)
Protected Methods
void	SetLayerParent (Network *parent)
bool	ConnectionInhibitory (float &pctInhibitory)
Protected Attributes
vector< Neuron * >	nrnLayer
Network *	parentNet
string	layerName
LayerType	layerType
unsigned int	layerId
float	learningConst
float	memTimeConst
float	synTimeConst
float	restPtnl
float	thresholdPtnl
bool	constantsSet
float	percentInhib
AmTimeInt	synapticDelay
Friends
class	Network

---

Detailed Description

Layer groups a number of Neurons for convenience. Layer is not needed to operate the network.

The only limitation currently is that unlayered Networks cannot be loaded or saved. This limitation should be addressed in 0.3.

See also:

Neuron, Network, NetLoader

Author:

Matt Grover

---

Member Function Documentation

void Layer::AddNeuron (  Neuron *    nrn )

Add a pre instantiated Neuron to the Layer. This doesn't affect the ownership of the Neuron until the Layer is added to a Network. At that point, the Network will assume ownership of the pointer. Neurons that are added to layers do not need to be added to a Network separately. Parameters: nrn  Pointer to a Neuron object that will be added to the Layer. See also: Network::AddLayer()

void Layer::AddNeuronVector (  vector< Neuron * >    nrnVec )

Add pre instantiated Neuron objects to the Layer. This doesn't affect the ownership of the Neuron until the Layer is added to a Network. At that point, the Network will assume ownership of the pointer. Neurons that are added to layers do not need to be added to a Network separately. Parameters: nrnVec  Vector of pointers to Neuron objects that will be added to the Layer. See also: Network::AddLayer()

const_iterator Layer::begin (    )  const [inline]

Returns: An iterator (vector<Neuron*>) pointing to the first neuron in the layer.

iterator Layer::begin (    )  [inline]

Returns: An iterator (vector<Neuron*>) pointing to the first neuron in the layer.

bool Layer::ConnectLayers (  Layer *    output, GaussConnectType    parms, float    pctInhibitory = 0.0 )

Connect this layer to to another layer. Parameters: output  Layer that will receive input from this layer. parms  Gaussian connection parameters. Layers connected with this parameter type will have weights set from a random Gaussian distribution. pctInhibitory  If Neuron::EnforceSign() has not been set prior to connecting the layers, then this can be used to set the number of connections to output that will have a negative weight. pctInhibitory will be ignored if sign enforcement has been turned on and must be in the range 0 -> 100.0. See also: Node::ConnectLayers()

bool Layer::ConnectLayers (  Layer *    output, UniConnectType    parms, float    pctInhibitory = 0.0 )

Connect this layer to to another layer. Parameters: output  Layer that will receive input from this layer. parms  Uniform connection parameters. Layers connected with this parameter type will have weights set from a uniform random distribution. pctInhibitory  If Neuron::EnforceSign() has not been set prior to connecting the layers, then this can be used to set the number of connections to output that will have a negative weight. pctInhibitory will be ignored if sign enforcement has been turned on and must be in the range 0 -> 100.0. See also: Node::ConnectLayers()

const_iterator Layer::end (    )  const [inline]

Returns: An iterator (vector<Neuron*>) pointing to the last neuron in the layer.

iterator Layer::end (    )  [inline]

Returns: An iterator (vector<Neuron*>) pointing to the last neuron in the layer.

LayerConstants Layer::GetLayerConstants (    )

Retrieve the set of constants that are common to all Neurons in a layer if they have been previously set. Returns: LayerConstants struct. See also: SetLayerConstants()

LayerType Layer::GetLayerType (    )  [inline]

Get the layer type Returns: layer type

AmTimeInt Layer::GetSynapticDelay (    )  const [inline]

Get the default spike transmission delay. Returns: The value of the delay in microseconds (us).

unsigned int Layer::LayerId (    )  [inline]

Get the numeric ID of the Layer. Returns: Numeric ID.

string Layer::LayerName (    )  [inline]

Retrieve the name of the Layer, Returns: Layer name.

void Layer::LayerName (  string    name )  [inline]

Set the name of the Layer. This is optional, but it can be useful after reloading a Network from a file. Parameters: name  Name of the Layer.

Neuron* Layer::operator[] (  unsigned int &    index )  [inline]

Returns: A pointer to a member Neuron. The index is not related to the NeuronID.

const_reverse_iterator Layer::rbegin (    )  const [inline]

Returns: A reverse iterator (vector<Neuron*>) pointing to the first neuron in the layer.

reverse_iterator Layer::rbegin (    )  [inline]

Returns: A reverse iterator (vector<Neuron*>) pointing to the first neuron in the layer.

const_reverse_iterator Layer::rend (    )  const [inline]

Returns: A reverse iterator (vector<Neuron*>) pointing to the last neuron in the layer.

reverse_iterator Layer::rend (    )  [inline]

Returns: A reverse iterator (vector<Neuron*>) pointing to the last neuron in the layer.

bool Layer::SetLayerConstants (  LayerConstants    lconst )

Set constants that should be common to all neurons in a layer, such as time constants and learning constants. This is optional and layers can be constructed without requiring that all member Neurons have the constants. Parameters: lconst  Layer constant structure. See also: GetLayerConstants()

void Layer::SetLayerId (  unsigned int    id )  [inline]

Set the unique numeric ID for the Layer. Parameters: id  Numeric id.

void Layer::SetLayerType (  LayerType    ltype )  [inline]

Designate what kind of layer this is ap ltype layer type.

void Layer::SetPercentInhibitory (  float    percent )

Set the percentage of neurons that should be inhibitory in the layer. This value can be overridden if it is also set in ConnectLayers if Neuron::EnforceSign() has not been called. Parameters: percent  Percentage of neurons that are inhibitory (0 - 100.0) See also: Neuron::EnforceSign()

void Layer::SetSynapticDelay (  AmTimeInt    delay )  [inline]

Set the default spike transmission delay. Parameters: delay  The value of the delay in microseconds (us).

unsigned int Layer::size (    )  [inline]

Returns: The size of the layer.

---

The documentation for this class was generated from the following file:

• layer.h

---