neupy.layers.pooling module

class neupy.layers.pooling.MaxPooling[source]

Maximum pooling layer.

Parameters:
size : tuple with 2 integers

Factor by which to downscale (vertical, horizontal). (2, 2) will halve the image in each dimension.

stride : tuple or int.

Stride size, which is the number of shifts over rows/cols to get the next pool region. If stride is None, it is considered equal to size (no overlap on pooling regions).

padding : {valid, same}

(pad_h, pad_w), pad zeros to extend beyond four borders of the images, pad_h is the size of the top and bottom margins, and pad_w is the size of the left and right margins.

name : str or None

Layer’s name. Can be used as a reference to specific layer. Name Can be specified as:

  • String: Specified name will be used as a direct reference to the layer. For example, name=”fc”
  • Format string: Name pattern could be defined as a format string and specified field will be replaced with an index. For example, name=”fc{}” will be replaced with fc1, fc2 and so on. A bit more complex formatting methods are acceptable, for example, name=”fc-{:<03d}” will be converted to fc-001, fc-002, fc-003 and so on.
  • None: When value specified as None than name will be generated from the class name.

Defaults to None.

Examples

2D pooling

>>> from neupy.layers import *
>>> network = join(
...     Input((10, 10, 3)),
...     MaxPooling((2, 2)),
... )
>>> network
(?, 10, 10, 3) -> [... 2 layers ...] -> (?, 5, 5, 3)

1D pooling

>>> from neupy.layers import *
>>> network = join(
...     Input((30, 10)),
...     Reshape((30, 1, 10)),
...     MaxPooling((2, 1)),
...     Reshape((-1, 10))
... )
>>> network
(?, 30, 10) -> [... 4 layers ...] -> (?, 15, 10)
Attributes:
variables : dict

Variable names and their values. Dictionary can be empty in case if variables hasn’t been created yet.

Methods

variable(value, name, shape=None, trainable=True) Initializes variable with specified values.
get_output_shape(input_shape) Computes expected output shape from the layer based on the specified input shape.
output(*inputs, **kwargs) Propagates input through the layer. The kwargs variable might contain additional information that propagates through the network.
options = {'name': Option(class_name='BaseLayer', value=Property(name="name")), 'padding': Option(class_name='BasePooling', value=ChoiceProperty(name="padding")), 'size': Option(class_name='BasePooling', value=TypedListProperty(name="size")), 'stride': Option(class_name='BasePooling', value=Spatial2DProperty(name="stride"))}[source]
pooling_type = 'MAX'[source]
class neupy.layers.pooling.AveragePooling[source]

Average pooling layer.

Parameters:
size : tuple with 2 integers

Factor by which to downscale (vertical, horizontal). (2, 2) will halve the image in each dimension.

stride : tuple or int.

Stride size, which is the number of shifts over rows/cols to get the next pool region. If stride is None, it is considered equal to size (no overlap on pooling regions).

padding : {valid, same}

(pad_h, pad_w), pad zeros to extend beyond four borders of the images, pad_h is the size of the top and bottom margins, and pad_w is the size of the left and right margins.

name : str or None

Layer’s name. Can be used as a reference to specific layer. Name Can be specified as:

  • String: Specified name will be used as a direct reference to the layer. For example, name=”fc”
  • Format string: Name pattern could be defined as a format string and specified field will be replaced with an index. For example, name=”fc{}” will be replaced with fc1, fc2 and so on. A bit more complex formatting methods are acceptable, for example, name=”fc-{:<03d}” will be converted to fc-001, fc-002, fc-003 and so on.
  • None: When value specified as None than name will be generated from the class name.

Defaults to None.

Examples

2D pooling

>>> from neupy.layers import *
>>> network = join(
...     Input((10, 10, 3)),
...     AveragePooling((2, 2)),
... )
>>> network
(?, 10, 10, 3) -> [... 2 layers ...] -> (?, 5, 5, 3)

1D pooling

>>> from neupy.layers import *
>>> network = join(
...     Input((30, 10)),
...     Reshape((30, 1, 10)),
...     AveragePooling((2, 1)),
...     Reshape((-1, 10))
... )
>>> network
(?, 30, 10) -> [... 4 layers ...] -> (?, 15, 10)
Attributes:
variables : dict

Variable names and their values. Dictionary can be empty in case if variables hasn’t been created yet.

Methods

variable(value, name, shape=None, trainable=True) Initializes variable with specified values.
get_output_shape(input_shape) Computes expected output shape from the layer based on the specified input shape.
output(*inputs, **kwargs) Propagates input through the layer. The kwargs variable might contain additional information that propagates through the network.
options = {'name': Option(class_name='BaseLayer', value=Property(name="name")), 'padding': Option(class_name='BasePooling', value=ChoiceProperty(name="padding")), 'size': Option(class_name='BasePooling', value=TypedListProperty(name="size")), 'stride': Option(class_name='BasePooling', value=Spatial2DProperty(name="stride"))}[source]
pooling_type = 'AVG'[source]
class neupy.layers.pooling.Upscale[source]

Upscales input over two axis (height and width).

Parameters:
scale : int or tuple with two int

Scaling factor for the input value. In the tuple first parameter identifies scale of the height and the second one of the width.

name : str or None

Layer’s name. Can be used as a reference to specific layer. Name Can be specified as:

  • String: Specified name will be used as a direct reference to the layer. For example, name=”fc”
  • Format string: Name pattern could be defined as a format string and specified field will be replaced with an index. For example, name=”fc{}” will be replaced with fc1, fc2 and so on. A bit more complex formatting methods are acceptable, for example, name=”fc-{:<03d}” will be converted to fc-001, fc-002, fc-003 and so on.
  • None: When value specified as None than name will be generated from the class name.

Defaults to None.

Examples

>>> from neupy.layers import *
>>> network = Input((10, 10, 3)) >> Upscale((2, 2))
(?, 10, 10, 3) -> [... 2 layers ...] -> (?, 20, 20, 3)
Attributes:
variables : dict

Variable names and their values. Dictionary can be empty in case if variables hasn’t been created yet.

Methods

variable(value, name, shape=None, trainable=True) Initializes variable with specified values.
get_output_shape(input_shape) Computes expected output shape from the layer based on the specified input shape.
output(*inputs, **kwargs) Propagates input through the layer. The kwargs variable might contain additional information that propagates through the network.
fail_if_shape_invalid(input_shape)[source]
get_output_shape(input_shape)[source]
options = {'name': Option(class_name='BaseLayer', value=Property(name="name")), 'scale': Option(class_name='Upscale', value=TypedListProperty(name="scale"))}[source]
output(input_value, **kwargs)[source]
scale = None[source]
class neupy.layers.pooling.GlobalPooling[source]

Global pooling layer.

Parameters:
function : {avg, max, sum} or callable

Common functions has been predefined for the user. These options are available:

  • avg - For average global pooling. The same as tf.reduce_mean.
  • max - For max global pooling. The same as tf.reduce_max.
  • sum - For sum global pooling. The same as tf.reduce_sum.

Parameter also excepts custom functions that have following format.

def agg_func(x, axis=None):
    pass

Defaults to avg.

name : str or None

Layer’s name. Can be used as a reference to specific layer. Name Can be specified as:

  • String: Specified name will be used as a direct reference to the layer. For example, name=”fc”
  • Format string: Name pattern could be defined as a format string and specified field will be replaced with an index. For example, name=”fc{}” will be replaced with fc1, fc2 and so on. A bit more complex formatting methods are acceptable, for example, name=”fc-{:<03d}” will be converted to fc-001, fc-002, fc-003 and so on.
  • None: When value specified as None than name will be generated from the class name.

Defaults to None.

Examples

>>> from neupy.layers import *
>>> network = Input((4, 4, 16)) >> GlobalPooling('avg')
(?, 4, 4, 16) -> [... 2 layers ...] -> (?, 16)
Attributes:
variables : dict

Variable names and their values. Dictionary can be empty in case if variables hasn’t been created yet.

Methods

variable(value, name, shape=None, trainable=True) Initializes variable with specified values.
get_output_shape(input_shape) Computes expected output shape from the layer based on the specified input shape.
output(*inputs, **kwargs) Propagates input through the layer. The kwargs variable might contain additional information that propagates through the network.
function = None[source]
get_output_shape(input_shape)[source]
options = {'function': Option(class_name='GlobalPooling', value=FunctionWithOptionsProperty(name="function")), 'name': Option(class_name='BaseLayer', value=Property(name="name"))}[source]
output(input_value, **kwargs)[source]