diff options
Diffstat (limited to 'dok')
-rw-r--r-- | dok/index.dok | 50 |
1 files changed, 50 insertions, 0 deletions
diff --git a/dok/index.dok b/dok/index.dok index d4a0160..baac983 100644 --- a/dok/index.dok +++ b/dok/index.dok @@ -1377,6 +1377,45 @@ output[i][j][k] = bias[k] * input[dW*(i-1)+s)][dH*(j-1)+t][l] </file> +==== VolumetricConvolution ==== +{{anchor:nn.VolumetricConvolution}} + +<file lua> +module = nn.VolumetricConvolution(nInputPlane, nOutputPlane, kT, kW, kH [, dT, dW, dH]) +</file> + +Applies a 3D convolution over an input image composed of several input planes. The ''input'' tensor in +''forward(input)'' is expected to be a 4D tensor (''nInputPlane x time x height x width''). + +The parameters are the following: + * ''nInputPlane'': The number of expected input planes in the image given into ''forward()''. + * ''nOutputPlane'': The number of output planes the convolution layer will produce. + * ''kT'': The kernel size of the convolution in time + * ''kW'': The kernel width of the convolution + * ''kH'': The kernel height of the convolution + * ''dT'': The step of the convolution in the time dimension. Default is ''1''. + * ''dW'': The step of the convolution in the width dimension. Default is ''1''. + * ''dH'': The step of the convolution in the height dimension. Default is ''1''. + +Note that depending of the size of your kernel, several (of the last) +columns or rows of the input image might be lost. It is up to the user to +add proper padding in images. + +If the input image is a 4D tensor ''nInputPlane x time x height x width'', the output image size +will be ''nOutputPlane x otime x owidth x oheight'' where +<file lua> +otime = (time - kT) / dT + 1 +owidth = (width - kW) / dW + 1 +oheight = (height - kH) / dH + 1 . +</file> + +The parameters of the convolution can be found in ''self.weight'' (Tensor of +size ''nOutputPlane x nInputPlane x kT x kH x kW'') and ''self.bias'' (Tensor of +size ''nOutputPlane''). The corresponding gradients can be found in +''self.gradWeight'' and ''self.gradBias''. + +</file> + ==== SpatialConvolutionMap ==== {{anchor:nn.SpatialConvolutionMap}} @@ -1436,6 +1475,17 @@ Applies 2D max-pooling operation in ''kWxkH'' regions by step size ''dWxdH'' steps. The number of output features is equal to the number of input planes. +==== VoulmetricMaxPooling ==== +{{anchor:nn.VolumetricMaxPooling}} + +<file lua> +module = nn.VolumetricMaxPooling(kT, kW, kH [, dT, dW, dH]) +</file> + +Applies 3D max-pooling operation in ''kTxkWxkH'' regions by step size +''dTxdWxdH'' steps. The number of output features is equal to the number of +input planes. + ==== SpatialSubSampling ==== {{anchor:nn.SpatialSubSampling}} |