deepity.ConvolutionalPCN.ConvolutionalPCN Class Reference

Inheritance diagram for deepity.ConvolutionalPCN.ConvolutionalPCN:

Public Member Functions
None	__init__ (self, *Layer|Activation architecture, tuple[int, int, int] input_shape, int batch_size)
"ConvolutionalPCN"	configure (self, float learning_rate=1e-6, float inference_rate=0.1, float precision_rate=0.0, float lmbda=1e-2)
None	set_learning_rate (self, float learning_rate)
None	set_inference_rate (self, float inference_rate)
None	set_precision_rate (self, float precision_rate)
None	set_lambda (self, float lmbda)
None	compile (self)
None	randomize_weights (self, Optional[int] seed=None, str dist="normal")
None	clamp_input (self, npt.NDArray[np.float32] X)
None	project_forward (self)
None	update_precision (self)
float	train_step (self, npt.NDArray[np.float32] X, npt.NDArray[np.float32] Y, int steps)
float	train_step_with_projection (self, npt.NDArray[np.float32] X, npt.NDArray[np.float32] Y, int steps)
npt.NDArray[np.float32]	predict (self, npt.NDArray[np.float32] X, int steps)
npt.NDArray[np.float32]	predict_with_projection (self, npt.NDArray[np.float32] X, int steps)
"ConvolutionalPCN"	fit (self, npt.NDArray[np.float32] X, npt.NDArray[np.float32] Y, int epochs, int steps, Optional[float] initial_lr=None, float decay_rate=1.0, bool shuffle=True)

Public Attributes
	architecture = architecture
	input_shape = input_shape
	layers

Protected Member Functions
None	_validate_architecture (self)
None	_build_backend (self)
None	_require_configured (self)

Protected Attributes
Optional[float]	_learning_rate = None
Optional[float]	_inference_rate = None
Optional[float]	_precision_rate = None
Optional[float]	_lambda = None

Detailed Description

Precision-weighted Convolutional Predictive Coding Network.

The architecture is declared using Convolution and Activation
objects (passed positionally, beginning with a Convolution), while
`input_shape` and optimization/inference hyperparameters are
supplied separately, see __init__() and configure().

Constructor & Destructor Documentation

__init__()

None deepity.ConvolutionalPCN.ConvolutionalPCN.__init__	(		self,
		*Layer | Activation	architecture,
		tuple[int, int, int]	input_shape,
		int	batch_size )

@param architecture Convolution/Activation objects describing the layer stack.
@param input_shape (channels, height, width) of a single input image.
@param batch_size Number of samples processed per train_step()/predict() call.

Member Function Documentation

_build_backend()

None deepity.ConvolutionalPCN.ConvolutionalPCN._build_backend ( self )

protected

_require_configured()

None deepity.ConvolutionalPCN.ConvolutionalPCN._require_configured ( self )

protected

_validate_architecture()

None deepity.ConvolutionalPCN.ConvolutionalPCN._validate_architecture ( self )

protected

clamp_input()

None deepity.ConvolutionalPCN.ConvolutionalPCN.clamp_input	(		self,
		npt.NDArray[np.float32]	X )

compile()

None deepity.ConvolutionalPCN.ConvolutionalPCN.compile

(

self

)

configure()

"ConvolutionalPCN" deepity.ConvolutionalPCN.ConvolutionalPCN.configure	(		self,
		float	learning_rate = 1e-6,
		float	inference_rate = 0.1,
		float	precision_rate = 0.0,
		float	lmbda = 1e-2 )

Configure and build the convolutional network. Must be called
once, before train_step()/predict().

@param learning_rate Learning rate for the forward weights.
@param inference_rate Inference/settling rate.
@param precision_rate Learning rate for precision weighting.
@param lmbda Regularization parameter.

fit()

"ConvolutionalPCN" deepity.ConvolutionalPCN.ConvolutionalPCN.fit	(		self,
		npt.NDArray[np.float32]	X,
		npt.NDArray[np.float32]	Y,
		int	epochs,
		int	steps,
		Optional[float]	initial_lr = None,
		float	decay_rate = 1.0,
		bool	shuffle = True )

Train for multiple epochs over (X, Y), batching by
`self.batch_size` and printing a progress bar.

predict()

npt.NDArray[np.float32] deepity.ConvolutionalPCN.ConvolutionalPCN.predict	(		self,
		npt.NDArray[np.float32]	X,
		int	steps )

Run inference on `X`, settling for `steps` iterations.
@return The terminal layer's beliefs.

predict_with_projection()

npt.NDArray[np.float32] deepity.ConvolutionalPCN.ConvolutionalPCN.predict_with_projection	(		self,
		npt.NDArray[np.float32]	X,
		int	steps )

Same as predict(), but initializes every layer's beliefs
via a forward projection pass first.

project_forward()

None deepity.ConvolutionalPCN.ConvolutionalPCN.project_forward

(

self

)

randomize_weights()

None deepity.ConvolutionalPCN.ConvolutionalPCN.randomize_weights	(		self,
		Optional[int]	seed = None,
		str	dist = "normal" )

Randomize all network weights. `seed` gives a reproducible run;
`dist` e.g. "normal(0,1)" or "uniform(-0.3,0.3)" (the default
"normal" has no "(a,b)" and so deliberately falls back to each
layer's own existing init).

set_inference_rate()

None deepity.ConvolutionalPCN.ConvolutionalPCN.set_inference_rate	(		self,
		float	inference_rate )

Set the inference/settling rate on every layer.

set_lambda()

None deepity.ConvolutionalPCN.ConvolutionalPCN.set_lambda	(		self,
		float	lmbda )

Set the regularization parameter on every layer.

set_learning_rate()

None deepity.ConvolutionalPCN.ConvolutionalPCN.set_learning_rate	(		self,
		float	learning_rate )

Set the learning rate on every layer.

set_precision_rate()

None deepity.ConvolutionalPCN.ConvolutionalPCN.set_precision_rate	(		self,
		float	precision_rate )

Set the precision-weighting learning rate on every layer.

train_step()

float deepity.ConvolutionalPCN.ConvolutionalPCN.train_step	(		self,
		npt.NDArray[np.float32]	X,
		npt.NDArray[np.float32]	Y,
		int	steps )

Run one training step on a single batch, settling for
`steps` iterations before updating weights.
@return Total settling energy accumulated over `steps`.

train_step_with_projection()

float deepity.ConvolutionalPCN.ConvolutionalPCN.train_step_with_projection	(		self,
		npt.NDArray[np.float32]	X,
		npt.NDArray[np.float32]	Y,
		int	steps )

Same as train_step(), but initializes every layer's
beliefs via a forward projection pass first instead of
whatever state they held from the previous call.

update_precision()

None deepity.ConvolutionalPCN.ConvolutionalPCN.update_precision

(

self

)

Member Data Documentation

_inference_rate

Optional[float] deepity.ConvolutionalPCN.ConvolutionalPCN._inference_rate = None

protected

_lambda

Optional[float] deepity.ConvolutionalPCN.ConvolutionalPCN._lambda = None

protected

_learning_rate

Optional[float] deepity.ConvolutionalPCN.ConvolutionalPCN._learning_rate = None

protected

_precision_rate

Optional[float] deepity.ConvolutionalPCN.ConvolutionalPCN._precision_rate = None

protected

architecture

deepity.ConvolutionalPCN.ConvolutionalPCN.architecture = architecture

input_shape

deepity.ConvolutionalPCN.ConvolutionalPCN.input_shape = input_shape

layers

deepity.ConvolutionalPCN.ConvolutionalPCN.layers

---

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

• deepity/ConvolutionalPCN.py