def max_pooling_2d(data, size: tuple, step: tuple, reference=None):
if reference is None:
from paradox.neural_network.convolutional_neural_network.operator import MaxPooling2D
return Symbol(operator=MaxPooling2D(size, step), inputs=as_symbols([data]), category=SymbolCategory.operator)
else:
from paradox.neural_network.convolutional_neural_network.operator import MaxReferencePooling2D
return Symbol(operator=MaxReferencePooling2D(size, step), inputs=as_symbols([data, reference]), category=SymbolCategory.operator)
def max_pooling_nd(data,
size: tuple,
step: tuple,
dimension: int,
reference=None):
if reference is None:
from paradox.neural_network.convolutional_neural_network.operator import MaxPoolingND
return Symbol(operator=MaxPoolingND(dimension, size, step),
inputs=as_symbols([data]))
else:
from paradox.neural_network.convolutional_neural_network.operator import MaxReferencePoolingND
return Symbol(operator=MaxReferencePoolingND(dimension, size, step),
inputs=as_symbols([data, reference]))
def convolution_nd(data,
kernel,
dimension: int,
mode,
element_wise: bool = False):
from paradox.neural_network.convolutional_neural_network.operator import ConvolutionND
return Symbol(operator=ConvolutionND(dimension, mode, element_wise),
inputs=as_symbols([data, kernel]))
def average_unpooling_nd(pooling,
size: tuple,
step: tuple,
dimension: int,
unpooling_size: int = None):
from paradox.neural_network.convolutional_neural_network.operator import AverageUnpoolingND
return Symbol(operator=AverageUnpoolingND(dimension, size, step,
unpooling_size),
inputs=as_symbols([pooling]))
def average_unpooling_2d(pooling, size: tuple, step: tuple, unpooling_size: tuple=None):
from paradox.neural_network.convolutional_neural_network.operator import AverageUnpooling2D
return Symbol(operator=AverageUnpooling2D(size, step, unpooling_size), inputs=as_symbols([pooling]), category=SymbolCategory.operator)
def average_pooling_2d(data, size: tuple, step: tuple):
from paradox.neural_network.convolutional_neural_network.operator import AveragePooling2D
return Symbol(operator=AveragePooling2D(size, step), inputs=as_symbols([data]), category=SymbolCategory.operator)
def max_unpooling_2d(data, pooling, size: tuple, step: tuple):
from paradox.neural_network.convolutional_neural_network.operator import MaxUnpooling2D
return Symbol(operator=MaxUnpooling2D(size, step), inputs=as_symbols([data, pooling]), category=SymbolCategory.operator)
def average_pooling_nd(data, size: tuple, step: tuple, dimension: int):
from paradox.neural_network.convolutional_neural_network.operator import AveragePoolingND
return Symbol(operator=AveragePoolingND(dimension, size, step),
inputs=as_symbols([data]))
def max_unpooling_nd(data, pooling, size: tuple, step: tuple, dimension: int):
from paradox.neural_network.convolutional_neural_network.operator import MaxUnpoolingND
return Symbol(operator=MaxUnpoolingND(dimension, size, step),
inputs=as_symbols([data, pooling]))