Convolutional PC layer, precision-free, AdamW-capable, routed through IComputeBackend for GPU portability – see the file-level warning above for the GPU path's verification status. More...

#include <SimpleConvPCLayer.h>

Inheritance diagram for Deep::SimpleConvPCLayer:

Public Member Functions
	SimpleConvPCLayer (int inChannels, int outChannels, int inHeight, int inWidth, int kernelH, int kernelW, int strideH=1, int strideW=1, int padH=0, int padW=0, int batchSize=1, float learningRate=1e-6f, float inferenceRate=0.1f, float lmbda=1e-2f, ActivationType aType=ActivationType::RELU, ActivationType dType=ActivationType::dRELU, IComputeBackend *backend=nullptr)
float	CalculateState () noexcept override
	Calculates the internal state of the layer.
void	UpdateState () noexcept override
	Updates the states after calculation.
void	UpdateWeights () noexcept override
	Updates the weights after state updates.
void	Flush () noexcept override
	Flushes remaining batches.
void	ClampState (const std::vector< float > &inputData) noexcept
	Clamps this layer's beliefs to inputData, fixing them against UpdateState() until UnclampState() is called.
void	UnclampState () noexcept
	Releases a previous ClampState() call.
float *	GetBeliefs () noexcept override
	Should return belief state z.
const float *	GetErrors () const noexcept override
	Should return errors e.
size_t	GetInputSize () const noexcept override
	Should return input size.
size_t	GetOutputSize () const noexcept override
	Should return output size.
size_t	GetBatchSize () const noexcept override
	Should return batch size.
const float *	GetWeights () const noexcept
	Returns the weight tensor, shape [outChannels, inChannelskernelHkernelW].
float *	GetWeights () noexcept
	Mutable overload of GetWeights() above.
const float *	GetBiases () const noexcept
	Returns the bias vector, length outChannels.
float *	GetBiases () noexcept
	Mutable overload of GetBiases() above.
float	GetLearningRate () const noexcept
	Returns the current learning rate.
float	GetInferenceRate () const noexcept
	Returns the current inference rate.
float	GetLambda () const noexcept
	Returns the current weight decay coefficient.
void	SetLearningRate (float lr) noexcept
	Sets the learning rate, propagating it to the optimizer state.
void	SetInferenceRate (float ir) noexcept
	Sets the inference rate.
void	SetLambda (float l) noexcept
	Sets the weight decay coefficient.
void	SetOptimizer (const OptimizerType o) noexcept
	Sets the weight optimizer. Call BEFORE Compile().
bool	IsClamped () const noexcept
	Whether ClampState() is currently active on this layer.
void	SetLayerAbove (SimpleConvPCLayer *above) noexcept
	Sets the layer immediately above this one in the network.
void	SetLayerBelow (SimpleConvPCLayer *below) noexcept
	Sets the layer immediately below this one in the network.
void	ComputeMuOnly () noexcept
	Computes only mu (forward prediction), skipping error/energy.
void	ResetState () noexcept
	Resets this layer's beliefs (z) without touching learned weights.
void	RandomizeWeights (std::mt19937 &twister) noexcept
	Randomizes W/b.
ActivationType	GetActivationType () const noexcept
	Returns this layer's activation type.
ActivationType	GetDerivativeType () const noexcept
	Returns this layer's activation derivative type.
const float *	GetMu () const noexcept
	Returns this layer's forward prediction buffer.
int	GetInChannels () const noexcept
	Returns the input channel count.
int	GetOutChannels () const noexcept
	Returns the output channel count (0 for a terminal layer).
int	GetInHeight () const noexcept
	Returns the input height.
int	GetInWidth () const noexcept
	Returns the input width.
int	GetOutHeight () const noexcept
	Returns the output height.
int	GetOutWidth () const noexcept
	Returns the output width.
int	GetKernelH () const noexcept
	Returns the kernel height.
int	GetKernelW () const noexcept
	Returns the kernel width.
size_t	GetRequiredFloats () const noexcept
	Total floats this layer needs from its MemoryArena.
template<typename ArenaT>
void	BindMemory (ArenaT &arena)
	Binds this layer's buffers into a pre-allocated arena (MemoryArena for CPU, DeviceMemoryArena for GPU).
Public Member Functions inherited from Deep::Layer
virtual	~Layer ()=default
virtual std::map< std::string, TensorDescriptor >	GetStateDict () const
	Returns a dictionary of this.

Friends
class	SimpleConvPCNDiagnostics

Additional Inherited Members
Protected Attributes inherited from Deep::Layer
size_t	size
	Size of input.
size_t	nextSize
	Size of output.

Detailed Description

Convolutional PC layer, precision-free, AdamW-capable, routed through IComputeBackend for GPU portability – see the file-level warning above for the GPU path's verification status.

Constructor & Destructor Documentation

◆ SimpleConvPCLayer()

Deep::SimpleConvPCLayer::SimpleConvPCLayer	(	int	inChannels,
		int	outChannels,
		int	inHeight,
		int	inWidth,
		int	kernelH,
		int	kernelW,
		int	strideH = 1,
		int	strideW = 1,
		int	padH = 0,
		int	padW = 0,
		int	batchSize = 1,
		float	learningRate = 1e-6f,
		float	inferenceRate = 0.1f,
		float	lmbda = 1e-2f,
		ActivationType	aType = ActivationType::RELU,
		ActivationType	dType = ActivationType::dRELU,
		IComputeBackend *	backend = nullptr )

Parameters

inChannels,outChannels	Input/output channel counts.
inHeight,inWidth	Input spatial dimensions.
kernelH,kernelW	Convolution kernel size.
strideH,strideW	Convolution stride.
padH,padW	Zero-padding on each spatial dimension.
batchSize	Batch size.
learningRate	Learning rate for W/b.
inferenceRate	Inference rate (Euler integration step size).
lmbda	Weight decay (L2 regularization) coefficient.
aType	Activation type.
dType	Activation derivative type.
backend	Compute backend to run on; defaults to a new CPUBackend if nullptr.

Member Function Documentation

◆ BindMemory()

template<typename ArenaT>

template void Deep::SimpleConvPCLayer::BindMemory< DeviceMemoryArena > ( ArenaT & arena )

Binds this layer's buffers into a pre-allocated arena (MemoryArena for CPU, DeviceMemoryArena for GPU).

Template Parameters

ArenaT Either MemoryArena or DeviceMemoryArena.

◆ CalculateState()

float Deep::SimpleConvPCLayer::CalculateState ( )

overridevirtualnoexcept

Calculates the internal state of the layer.

Returns: energy used

Implements Deep::Layer.

◆ ClampState()

void Deep::SimpleConvPCLayer::ClampState ( const std::vector< float > & inputData )

noexcept

Clamps this layer's beliefs to inputData, fixing them against UpdateState() until UnclampState() is called.

◆ ComputeMuOnly()

void Deep::SimpleConvPCLayer::ComputeMuOnly ( )

noexcept

Computes only mu (forward prediction), skipping error/energy.

◆ Flush()

void Deep::SimpleConvPCLayer::Flush ( )

inlineoverridevirtualnoexcept

Flushes remaining batches.

Reimplemented from Deep::Layer.

◆ GetActivationType()

ActivationType Deep::SimpleConvPCLayer::GetActivationType ( ) const

inlinenoexcept

Returns this layer's activation type.

◆ GetBatchSize()

size_t Deep::SimpleConvPCLayer::GetBatchSize ( ) const

inlineoverridevirtualnoexcept

Should return batch size.

Returns: ? = 0

Implements Deep::Layer.

◆ GetBeliefs()

float * Deep::SimpleConvPCLayer::GetBeliefs ( )

inlineoverridevirtualnoexcept

Should return belief state z.

Returns: ? = 0

Implements Deep::Layer.

◆ GetBiases() [1/2]

const float * Deep::SimpleConvPCLayer::GetBiases ( ) const

inlinenoexcept

Returns the bias vector, length outChannels.

◆ GetBiases() [2/2]

float * Deep::SimpleConvPCLayer::GetBiases ( )

inlinenoexcept

Mutable overload of GetBiases() above.

◆ GetDerivativeType()

ActivationType Deep::SimpleConvPCLayer::GetDerivativeType ( ) const

inlinenoexcept

Returns this layer's activation derivative type.

◆ GetErrors()

const float * Deep::SimpleConvPCLayer::GetErrors ( ) const

inlineoverridevirtualnoexcept

Should return errors e.

Returns: ? = 0

Implements Deep::Layer.

◆ GetInChannels()

int Deep::SimpleConvPCLayer::GetInChannels ( ) const

inlinenoexcept

Returns the input channel count.

◆ GetInferenceRate()

float Deep::SimpleConvPCLayer::GetInferenceRate ( ) const

inlinenoexcept

Returns the current inference rate.

◆ GetInHeight()

int Deep::SimpleConvPCLayer::GetInHeight ( ) const

inlinenoexcept

Returns the input height.

◆ GetInputSize()

size_t Deep::SimpleConvPCLayer::GetInputSize ( ) const

inlineoverridevirtualnoexcept

Should return input size.

Returns: ? = 0

Implements Deep::Layer.

◆ GetInWidth()

int Deep::SimpleConvPCLayer::GetInWidth ( ) const

inlinenoexcept

Returns the input width.

◆ GetKernelH()

int Deep::SimpleConvPCLayer::GetKernelH ( ) const

inlinenoexcept

Returns the kernel height.

◆ GetKernelW()

int Deep::SimpleConvPCLayer::GetKernelW ( ) const

inlinenoexcept

Returns the kernel width.

◆ GetLambda()

float Deep::SimpleConvPCLayer::GetLambda ( ) const

inlinenoexcept

Returns the current weight decay coefficient.

◆ GetLearningRate()

float Deep::SimpleConvPCLayer::GetLearningRate ( ) const

inlinenoexcept

Returns the current learning rate.

◆ GetMu()

const float * Deep::SimpleConvPCLayer::GetMu ( ) const

inlinenoexcept

Returns this layer's forward prediction buffer.

◆ GetOutChannels()

int Deep::SimpleConvPCLayer::GetOutChannels ( ) const

inlinenoexcept

Returns the output channel count (0 for a terminal layer).

◆ GetOutHeight()

int Deep::SimpleConvPCLayer::GetOutHeight ( ) const

inlinenoexcept

Returns the output height.

◆ GetOutputSize()

size_t Deep::SimpleConvPCLayer::GetOutputSize ( ) const

inlineoverridevirtualnoexcept

Should return output size.

Returns: ? = 0

Implements Deep::Layer.

◆ GetOutWidth()

int Deep::SimpleConvPCLayer::GetOutWidth ( ) const

inlinenoexcept

Returns the output width.

◆ GetRequiredFloats()

size_t Deep::SimpleConvPCLayer::GetRequiredFloats ( ) const

noexcept

Total floats this layer needs from its MemoryArena.

◆ GetWeights() [1/2]

const float * Deep::SimpleConvPCLayer::GetWeights ( ) const

inlinenoexcept

Returns the weight tensor, shape [outChannels, inChannels*kernelH*kernelW].

◆ GetWeights() [2/2]

float * Deep::SimpleConvPCLayer::GetWeights ( )

inlinenoexcept

Mutable overload of GetWeights() above.

◆ IsClamped()

bool Deep::SimpleConvPCLayer::IsClamped ( ) const

inlinenoexcept

Whether ClampState() is currently active on this layer.

◆ RandomizeWeights()

void Deep::SimpleConvPCLayer::RandomizeWeights ( std::mt19937 & twister )

noexcept

Randomizes W/b.

◆ ResetState()

void Deep::SimpleConvPCLayer::ResetState ( )

noexcept

Resets this layer's beliefs (z) without touching learned weights.

◆ SetInferenceRate()

void Deep::SimpleConvPCLayer::SetInferenceRate ( float ir )

inlinenoexcept

Sets the inference rate.

◆ SetLambda()

void Deep::SimpleConvPCLayer::SetLambda ( float l )

inlinenoexcept

Sets the weight decay coefficient.

◆ SetLayerAbove()

void Deep::SimpleConvPCLayer::SetLayerAbove ( SimpleConvPCLayer * above )

inlinenoexcept

Sets the layer immediately above this one in the network.

◆ SetLayerBelow()

void Deep::SimpleConvPCLayer::SetLayerBelow ( SimpleConvPCLayer * below )

inlinenoexcept

Sets the layer immediately below this one in the network.

◆ SetLearningRate()

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

noexcept

Sets the learning rate, propagating it to the optimizer state.

◆ SetOptimizer()

void Deep::SimpleConvPCLayer::SetOptimizer ( const OptimizerType o )

inlinenoexcept

Sets the weight optimizer. Call BEFORE Compile().

◆ UnclampState()

void Deep::SimpleConvPCLayer::UnclampState ( )

noexcept

Releases a previous ClampState() call.

◆ UpdateState()

void Deep::SimpleConvPCLayer::UpdateState ( )

overridevirtualnoexcept

Updates the states after calculation.

Implements Deep::Layer.

◆ UpdateWeights()

void Deep::SimpleConvPCLayer::UpdateWeights ( )

overridevirtualnoexcept

Updates the weights after state updates.

Implements Deep::Layer.

◆ SimpleConvPCNDiagnostics

friend class SimpleConvPCNDiagnostics

friend

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

include/deepity/layers/SimpleConvPCLayer.h
src/layers/SimpleConvPCLayer.cpp

Public Member Functions

Friends

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ SimpleConvPCLayer()

Member Function Documentation

◆ BindMemory()

◆ CalculateState()

◆ ClampState()

◆ ComputeMuOnly()

◆ Flush()

◆ GetActivationType()

◆ GetBatchSize()

◆ GetBeliefs()

◆ GetBiases() [1/2]

◆ GetBiases() [2/2]

◆ GetDerivativeType()

◆ GetErrors()

◆ GetInChannels()

◆ GetInferenceRate()

◆ GetInHeight()

◆ GetInputSize()

◆ GetInWidth()

◆ GetKernelH()

◆ GetKernelW()

◆ GetLambda()

◆ GetLearningRate()

◆ GetMu()

◆ GetOutChannels()

◆ GetOutHeight()

◆ GetOutputSize()

◆ GetOutWidth()

◆ GetRequiredFloats()

◆ GetWeights() [1/2]

◆ GetWeights() [2/2]

◆ IsClamped()

◆ RandomizeWeights()

◆ ResetState()

◆ SetInferenceRate()

◆ SetLambda()

◆ SetLayerAbove()

◆ SetLayerBelow()

◆ SetLearningRate()

◆ SetOptimizer()

◆ UnclampState()

◆ UpdateState()

◆ UpdateWeights()

Friends And Related Symbol Documentation

◆ SimpleConvPCNDiagnostics