Skip to main content
. 2019 Nov 12;9:16590. doi: 10.1038/s41598-019-53091-8

Figure 3.

Figure 3

The architectures of deep neural networks. (A) Inception module using 5 × 5, 3 × 3, and 1 × 1 convolution. It replaces the 5 × 5 convolution layer. (B) Inception module using 3 × 3 and 1 × 1 convolution. It replaces the 3 × 3 convolution layer. (C) Inception module using 1 × 7, 7 × 1, and 1 × 1 convolution. It replaces the 7 × 7 convolution layer. (D) Inception module using 3 × 3, 1 × 7, 7 × 1, and 1 × 1 convolution. It replaces the 7 × 7 convolution layer. (E) Inception module using 3 × 3, 1 × 3, 3 × 1, and 1 × 1 convolution. It replaces 3 × 3 convolution layer.