Skip to main content

Table 7 Layer structure of the LeNet-5 “deep learning” CNN employed in this investigation. Sizes refer to pixels for layers 1–7, variables for layers 8–10

From: Machine-learning strategies for testing patterns of morphological variation in small samples: sexual dimorphism in gray wolf (Canis lupus) crania

Layers

Type

Parameters

  

Image

1

Input

3-tensor (size 1 × 28 × 28)

2

Convolution

3-tensor (size 10 × 25 × 25)

3

Ramp

3-tensor (size 10 × 25 × 25)

4

Pooling

3-tensor (size 10 × 12 × 12)

5

Convolution

3-tensor (size 20 × 9 × 9)

6

Ramp

3-tensor (size 20 × 9 × 9)

7

Pooling

3-tensor (size 20 × 4 × 4)

8

Flatten

Vector (size 320)

9

Linear

Vector (size 2)

10

Output

Vector (size 2)