Deep::GaussSeidelPCNetwork Class Reference

Predictive Coding network built from GaussSeidelPCLayer, sweeping layers sequentially rather than updating them simultaneously. See the file-level warning above on its current verification status. More...

#include <GaussSeidelPCNetwork.h>

Public Member Functions
	GaussSeidelPCNetwork (int batchSize) noexcept
	Constructs an empty network with a predetermined batch size.
	~GaussSeidelPCNetwork ()=default
	Default destructor.
	GaussSeidelPCNetwork (const GaussSeidelPCNetwork &)=delete
	Copy construction is disabled; a network owns its layers' memory arena and is not meant to be duplicated.
GaussSeidelPCNetwork &	operator= (const GaussSeidelPCNetwork &)=delete
	Copy assignment is disabled; see the copy constructor.
void	AddLayer (int size, int nextSize, float lr, float ir, float lmbda, void(*act)(float *, size_t), void(*dAct)(float *, size_t, bool))
	Adds a layer to the network.
void	AddLayer (int size, int nextSize, float lr, float ir, float lmbda, ActivationType aType, ActivationType dType)
	Adds a layer to the network, using a named ActivationType instead of raw function pointers.
void	RandomizeWeights (std::mt19937 &rng)
	Randomizes the weights of each layer.
void	ResetState () noexcept
	Resets each layer's state without touching learned weights.
void	Clamp (const std::vector< float > &input)
	Clamps the input to the first layer.
float	Step () noexcept
	Runs ONE full Gauss-Seidel settling step (all three sweeps, in order) across every layer.
void	UpdateWeights () noexcept
	Updates every non-terminal layer's weights. Called ONCE after the full settling loop completes, matching ngc-learn's evolve_process.
void	ProjectForward () noexcept
	Seeds every hidden layer's z from a genuine forward pass through current weights (ComputePrediction() only, no error/energy side effects) – an orthogonal optimization to the Gauss-Seidel restructuring itself. Assumes layers[0] is already clamped.
void	SetOptimizer (OptimizerType o) noexcept
	Sets the optimizer used for weight updates, on every layer.
void	SetLearningRate (float lr) noexcept
	Sets the learning rate used for weight updates, on every layer.
GaussSeidelPCLayer *	GetTerminalLayer () noexcept
	Returns the network's terminal (final) layer.
std::vector< std::unique_ptr< GaussSeidelPCLayer > > &	GetLayers () noexcept
	Returns every layer in the network, in the order they were added.
const std::vector< std::unique_ptr< GaussSeidelPCLayer > > &	GetLayers () const noexcept
	Returns every layer in the network, in the order they were added.
int	GetBatchSize () const noexcept
	Returns the batch size for the network's layers.
float	TrainStep (const std::vector< float > &x, const std::vector< float > &y, int inferenceSteps)
	Full train step: reset, clamp input+target, settle for inferenceSteps, update weights once, unclamp.
float	TrainStepWithProjection (const std::vector< float > &x, const std::vector< float > &y, int inferenceSteps)
	Same as TrainStep(), but with ProjectForward() called once after clamping the input, before the settling loop.
std::vector< float >	Predict (const std::vector< float > &x, int inferenceSteps)
	Forward prediction: clamp input, settle (no target clamped), read the terminal's beliefs.
void	Compile ()
	Loads all layers into one contiguous block of memory.

Detailed Description

Predictive Coding network built from GaussSeidelPCLayer, sweeping layers sequentially rather than updating them simultaneously. See the file-level warning above on its current verification status.

Constructor & Destructor Documentation

GaussSeidelPCNetwork() [1/2]

Deep::GaussSeidelPCNetwork::GaussSeidelPCNetwork ( int batchSize )

explicitnoexcept

Constructs an empty network with a predetermined batch size.

Parameters

batchSize

Batch size

~GaussSeidelPCNetwork()

Deep::GaussSeidelPCNetwork::~GaussSeidelPCNetwork ( )

default

Default destructor.

GaussSeidelPCNetwork() [2/2]

Deep::GaussSeidelPCNetwork::GaussSeidelPCNetwork ( const GaussSeidelPCNetwork & )

delete

Copy construction is disabled; a network owns its layers' memory arena and is not meant to be duplicated.

Member Function Documentation

AddLayer() [1/2]

void Deep::GaussSeidelPCNetwork::AddLayer	(	int	size,
		int	nextSize,
		float	lr,
		float	ir,
		float	lmbda,
		ActivationType	aType,
		ActivationType	dType )

Adds a layer to the network, using a named ActivationType instead of raw function pointers.

Parameters

size	input size
nextSize	output size
lr	learning rate for weight updates
ir	inference rate (Euler integration step size)
lmbda	weight decay (L2 regularization) coefficient
aType	activation type
dType	activation derivative type

AddLayer() [2/2]

void Deep::GaussSeidelPCNetwork::AddLayer	(	int	size,
		int	nextSize,
		float	lr,
		float	ir,
		float	lmbda,
		void(*	act )(float *, size_t),
		void(*	dAct )(float *, size_t, bool) )

Adds a layer to the network.

Parameters

size	input size
nextSize	output size
lr	learning rate for weight updates
ir	inference rate (Euler integration step size)
lmbda	weight decay (L2 regularization) coefficient
act	activation function
dAct	derivative of previous activation function

Clamp()

void Deep::GaussSeidelPCNetwork::Clamp

(

const std::vector< float > &

input

)

Clamps the input to the first layer.

Parameters

input

Reference to the input vector.

Compile()

void Deep::GaussSeidelPCNetwork::Compile

(

)

Loads all layers into one contiguous block of memory.

GetBatchSize()

int Deep::GaussSeidelPCNetwork::GetBatchSize ( ) const

inlinenoexcept

Returns the batch size for the network's layers.

Returns

int batchSize

GetLayers() [1/2]

const std::vector< std::unique_ptr< GaussSeidelPCLayer > > & Deep::GaussSeidelPCNetwork::GetLayers ( ) const

inlinenoexcept

Returns every layer in the network, in the order they were added.

Returns

A const reference to the internal layer list.

GetLayers() [2/2]

std::vector< std::unique_ptr< GaussSeidelPCLayer > > & Deep::GaussSeidelPCNetwork::GetLayers ( )

inlinenoexcept

Returns every layer in the network, in the order they were added.

Returns

A reference to the internal layer list.

GetTerminalLayer()

GaussSeidelPCLayer * Deep::GaussSeidelPCNetwork::GetTerminalLayer ( )

inlinenoexcept

Returns the network's terminal (final) layer.

Returns

Pointer to the last layer added via AddLayer().

operator=()

GaussSeidelPCNetwork & Deep::GaussSeidelPCNetwork::operator= ( const GaussSeidelPCNetwork & )

delete

Copy assignment is disabled; see the copy constructor.

Predict()

std::vector< float > Deep::GaussSeidelPCNetwork::Predict	(	const std::vector< float > &	x,
		int	inferenceSteps )

Forward prediction: clamp input, settle (no target clamped), read the terminal's beliefs.

ProjectForward()

void Deep::GaussSeidelPCNetwork::ProjectForward ( )

noexcept

Seeds every hidden layer's z from a genuine forward pass through current weights (ComputePrediction() only, no error/energy side effects) – an orthogonal optimization to the Gauss-Seidel restructuring itself. Assumes layers[0] is already clamped.

RandomizeWeights()

void Deep::GaussSeidelPCNetwork::RandomizeWeights

(

std::mt19937 &

rng

)

Randomizes the weights of each layer.

Parameters

rng	The classic Mersenne Twister

ResetState()

void Deep::GaussSeidelPCNetwork::ResetState ( )

noexcept

Resets each layer's state without touching learned weights.

SetLearningRate()

void Deep::GaussSeidelPCNetwork::SetLearningRate ( float lr )

inlinenoexcept

Sets the learning rate used for weight updates, on every layer.

Parameters

lr	The new learning rate.

SetOptimizer()

void Deep::GaussSeidelPCNetwork::SetOptimizer ( OptimizerType o )

inlinenoexcept

Sets the optimizer used for weight updates, on every layer.

Parameters

o	The optimizer type to apply.

Step()

float Deep::GaussSeidelPCNetwork::Step ( )

noexcept

Runs ONE full Gauss-Seidel settling step (all three sweeps, in order) across every layer.

Returns

Total energy, summed from every layer's ComputeError() – meaningful only after all three sweeps have run for this step.

TrainStep()

float Deep::GaussSeidelPCNetwork::TrainStep	(	const std::vector< float > &	x,
		const std::vector< float > &	y,
		int	inferenceSteps )

Full train step: reset, clamp input+target, settle for inferenceSteps, update weights once, unclamp.

Returns

The final step's total energy, before weight updates.

TrainStepWithProjection()

float Deep::GaussSeidelPCNetwork::TrainStepWithProjection	(	const std::vector< float > &	x,
		const std::vector< float > &	y,
		int	inferenceSteps )

Same as TrainStep(), but with ProjectForward() called once after clamping the input, before the settling loop.

UpdateWeights()

void Deep::GaussSeidelPCNetwork::UpdateWeights ( )

noexcept

Updates every non-terminal layer's weights. Called ONCE after the full settling loop completes, matching ngc-learn's evolve_process.

---

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

• include/deepity/networks/GaussSeidelPCNetwork.h
• src/networks/GaussSeidelPCNetwork.cpp