Un-pooling is an operation used in convolutional generative networks to increase the spatial size of a representation, addressing the information loss caused by standard pooling layers. It operates under specific simplifying assumptions: 1. The stride of the max-pooling operation is constrained to equal the width of the pooling region. 2. The maximum input within each pooling region is assumed to be in the upper-left corner. 3. All non-maximal inputs within each pooling region are assumed to be zero. The inverse un-pooling operation allocates a tensor of zeros, then copies each value from spatial coordinate $i$ of the input to spatial coordinate $i\times k$ of the output, where $k$ is an integer defining the size of the pooling region.