Pooling

MaxPooling1D

layer eddl::MaxPool1D(layer parent, vector<int> pool_size = {2}, vector<int> strides = {2}, string padding = "none", string name = "")

MaxPooling1D operation.

Parameters

parent – Parent layer
pool_size – Size of the max pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after applying the max pooling operation over the parent layer

Example:

l = MaxPool1D(l, {4}, {4}, "same");

MaxPooling2D

layer eddl::MaxPool2D(layer parent, vector<int> pool_size = {2, 2}, vector<int> strides = {2, 2}, string padding = "none", string name = "")

MaxPooling2D operation.

Parameters

parent – Parent layer
pool_size – Size of the max pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after applying the max pooling operation over the parent layer

Example:

l = MaxPool2D(l, {3, 3}, {1, 1}, "same");

MaxPooling3D

layer eddl::MaxPool3D(layer parent, vector<int> pool_size = {2, 2, 2}, vector<int> strides = {2, 2, 2}, string padding = "none", string name = "")

MaxPooling3D operation.

Parameters

parent – Parent layer
pool_size – Size of the max pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after applying the max pooling operation over the parent layer

l = MaxPool3D(l, {3, 3, 3}, {3, 3, 3}, "same");

AveragePooling1D

layer eddl::AveragePool1D(layer parent, vector<int> pool_size = {2}, vector<int> strides = {2}, string padding = "none", string name = "")

AveragePooling1D operation.

Parameters

parent – Parent layer
pool_size – Size of the average pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after apply the average pooling operation over the parent layer

Example:

l = AveragePool1D(l, {4}, {4}, "same");

AveragePooling2D

layer eddl::AveragePool2D(layer parent, vector<int> pool_size = {2, 2}, vector<int> strides = {2, 2}, string padding = "none", string name = "")

AveragePooling2D operation.

Parameters

parent – Parent layer
pool_size – Size of the average pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after apply the average pooling operation over the parent layer

Example:

l = AveragePool2D(l, {3, 3}, {1, 1}, "same");

AveragePooling3D

layer eddl::AveragePool3D(layer parent, vector<int> pool_size = {2, 2, 2}, vector<int> strides = {2, 2, 2}, string padding = "none", string name = "")

AveragePooling3D operation.

Parameters

parent – Parent layer
pool_size – Size of the average pooling windows
strides – Factor by which to downscale. E.g. 2 will halve the input. If None, it will default to pool_size
padding – One of “none”, “valid” or “same” (case-insensitive)
name – A name for the operation

Returns

The result after apply the average pooling operation over the parent layer

l = AveragePool3D(l, {3, 3, 3}, {3, 3, 3}, "same");

GlobalMaxPooling1D

layer eddl::GlobalMaxPool1D(layer parent, string name = "")

GlobalMaxPooling1D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global max pooling operation over the parent layer

Example:

l = GlobalMaxPool1D(l);

GlobalMaxPooling2D

layer eddl::GlobalMaxPool2D(layer parent, string name = "")

GlobalMaxPooling2D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global max pooling operation over the parent layer

Example:

l = GlobalMaxPool2D(l);

GlobalMaxPooling3D

layer eddl::GlobalMaxPool3D(layer parent, string name = "")

GlobalMaxPooling3D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global max pooling operation over the parent layer

l = GlobalMaxPool3D(l);

GlobalAveragePooling1D

layer eddl::GlobalAveragePool1D(layer parent, string name = "")

GlobalAveragePooling1D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global average pooling operation over the parent layer

Example:

l = GlobalAveragePool1D(l);

GlobalAveragePooling2D

layer eddl::GlobalAveragePool2D(layer parent, string name = "")

GlobalAveragePooling2D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global average pooling operation over the parent layer

Example:

l = GlobalAveragePool2D(l);

GlobalAveragePooling3D

layer eddl::GlobalAveragePool3D(layer parent, string name = "")

GlobalAveragePooling3D operation.

Parameters

parent – Parent layer
name – A name for the operation

Returns

The result after applying the global average pooling operation over the parent layer

l = GlobalAveragePool3D(l);