Skip to main content
. 2022 Dec 14;44(4):1779–1792. doi: 10.1002/hbm.26174

FIGURE 7.

FIGURE 7

We split the 3D convolutional kernels into multiple 2D slices for visualization. (a) The first and second rows are the average magnitude matrixes of kernel size of 3 for convolutional layer in MixACB and conventional convolutional layer, respectively. (b) The first and second rows indicate the average magnitude matrixes of kernel size of 5 for convolutional layer in MixACB and conventional convolutional layer, respectively.