Neuron Class Reference

Represents a single neuron within a neural network layer. More...

#include <Neuron.hpp>

Public Member Functions
	Neuron (unsigned numOutputs, unsigned myIdx)
	Constructs a Neuron with a specified number of outputs. More...
void	setOutputVal (double val)
	Sets the neuron's output value. More...
double	getOutputVal () const
	Retrieves the neuron's output value. More...
void	feedForward (const Layer &prevLayer)
	Feeds the input values forward through the neuron. More...
void	calcOutputGradients (double targetVal)
	Calculates the gradient for output layer neurons. More...
void	calcHiddenGradients (const Layer &nextLayer)
	Calculates gradients for hidden layer neurons. More...
void	updateInputWeights (Layer &prevLayer)
	Updates the weights of connections from the previous layer. More...

Detailed Description

Represents a single neuron within a neural network layer.

The Neuron class handles the neuron's output value, weights for connections to subsequent neurons, and methods for updating weights and calculating gradients.

Constructor & Destructor Documentation

Neuron()

Neuron::Neuron	(	unsigned	numOutputs,
		unsigned	myIdx
	)

Constructs a Neuron with a specified number of outputs.

Constructor for the Neuron class.

Initializes a neuron with a certain number of connections (outputs) and an index within its layer.

Parameters

numOutputs	The number of connections from this neuron to the next layer.
myIdx	The index of this neuron in its layer.

Initializes a neuron with a specified number of outputs and index. Each output connection is initialized with a random weight.

Parameters

numOutputs	The number of connections from this neuron to the next layer.
myIdx	The index of this neuron in its layer.

Member Function Documentation

calcHiddenGradients()

void Neuron::calcHiddenGradients

(

const Layer &

nextLayer

)

Calculates gradients for hidden layer neurons.

Computes the gradient for hidden neurons based on the sum of weights and gradients from the next layer and the derivative of the transfer function.

Parameters

nextLayer

The next layer of neurons, used to compute the gradient.

calcOutputGradients()

void Neuron::calcOutputGradients

(

double

targetVal

)

Calculates the gradient for output layer neurons.

Uses the difference between the target and actual output values, adjusted by the derivative of the transfer function, to set the gradient.

Parameters

targetVal

The target value for the output neuron.

feedForward()

void Neuron::feedForward

(

const Layer &

prevLayer

)

Feeds the input values forward through the neuron.

Calculates the neuron's output by summing the products of inputs and weights, then applies the transfer function.

Parameters

prevLayer

The previous layer of neurons, used as inputs to this neuron.

Calculates the neuron's output by summing the products of inputs and weights, and then applies the transfer function to the result.

Parameters

prevLayer

The previous layer of neurons, used as inputs to this neuron.

getOutputVal()

double Neuron::getOutputVal ( ) const

inline

Retrieves the neuron's output value.

Returns

The output value of the neuron.

setOutputVal()

void Neuron::setOutputVal ( double  val )

inline

Sets the neuron's output value.

Parameters

val	The output value to be set.

updateInputWeights()

void Neuron::updateInputWeights

(

Layer &

prevLayer

)

Updates the weights of connections from the previous layer.

Adjusts each weight based on the neuron's gradient and learning parameters (eta and alpha).

Parameters

prevLayer

The layer of neurons from which the current neuron receives inputs.

---

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

• OrderbookSim/Neuron.hpp
• OrderbookSim/Neuron.cpp