|
|
Represents one (fully-connected) layer from a multi-layer percetron (FFNet)
|
Activation function pointer
|
Activation function derivative pointer
int |
[protected]
Number of neurons in the layer
int |
[protected]
Number of input neurons
float * |
[protected]
Pointer to the weight vector
string |
[protected]
int |
[protected]
Offset of the layer weight vector in the network weight vector
int |
[protected]
Offset of the layer's first neuron in the list of all network neurons (not too sure about that one, though)
float |
[protected]
Offset of the layer's first neuron derivative in the list of all network neurons
|
This (empty) constructor is used for parsing a layer from a file
|
Standard constructor
|
Unimplemented yet (not sure if we *should* implement it)
void |
Called after reading a layer to setup the weight vector correctly
~ |
void |
Calculates all the activation functions (and derivatives) for a given input
int |
Layer size
int |
Number of weights
int |
Offset of a certain neuron's weight vector relative to the whole network's weight vector
int |
int |
void |
void |
Initializes the layer values given the means and standard deviations of each of the inputs
void |
Sets the bias vector
float * |
void |
[const]
Writes the layer to a stream
Reimplemented from Object.
void |
Reads the layer from a stream (remember to call setupAfterRead after that
Reimplemented from Object.
void |