torch.nn.functional.fractional_max_pool3d

torch.nn.functional.fractional_max_pool3d(input, kernel_size, output_size=None, output_ratio=None, return_indices=False, _random_samples=None)

Applies 3D fractional max pooling over an input signal composed of several input planes.

Fractional MaxPooling is described in detail in the paper Fractional MaxPooling by Ben Graham

The max-pooling operation is applied in $kT \times kH \times kW$ regions by a stochastic step size determined by the target output size. The number of output features is equal to the number of input planes.

Parameters:

• kernel_size – the size of the window to take a max over. Can be a single number $k$ (for a square kernel of $k \times k \times k$) or a tuple (kT, kH, kW)

• output_size – the target output size of the form $oT \times oH \times oW$. Can be a tuple (oT, oH, oW) or a single number $oH$ for a cubic output $oH \times oH \times oH$

• output_ratio – If one wants to have an output size as a ratio of the input size, this option can be given. This has to be a number or tuple in the range (0, 1)

• return_indices – if True, will return the indices along with the outputs. Useful to pass to max_unpool3d().

Shape:

• Input: $(N, C, T_{in}, H_{in}, W_{in})$ or $(C, T_{in}, H_{in}, W_{in})$.

• Output: $(N, C, T_{out}, H_{out}, W_{out})$ or $(C, T_{out}, H_{out}, W_{out})$, where $(T_{out}, H_{out}, W_{out})=\text{output\_size}$ or $(T_{out}, H_{out}, W_{out})=\text{output\_ratio} \times (T_{in}, H_{in}, W_{in})$

Examples::

>>> input = torch.randn(20, 16, 50, 32, 16)
>>> # pool of cubic window of size=3, and target output size 13x12x11
>>> F.fractional_max_pool3d(input, 3, output_size=(13, 12, 11))
>>> # pool of cubic window and target output size being half of input size
>>> F.fractional_max_pool3d(input, 3, output_ratio=(0.5, 0.5, 0.5))