The proposed algorithm can also process input feature maps and generate output feature maps with the same flexible block sizes that are independent of convolution weight kernel size. The memory access ...

Chai et al. (2020, 2021) extended the 2D seismic reconstruction task to 3D by using an ordinary 3D Unet and obtained good interpolation results. Kong et al. (2022) proposed a multi-resolution Unet ...

Sparse convolution in python. Uses Toeplitz convolutional matrix multiplication to perform sparse convolution. This allows for extremely fast convolution when: The kernel is small (<= 100x100) The ...

To conclude that part, these 4 examples show that DiAGAN can generate 3D realizations that are close to the 3D references from 2D examples. There are some differences but one has to remember that the ...