Classroom Work
Q1: If you have 10 filters that are 3x3x3 in one layer of a convolutional neural network, how
many parameters does that layer have?
Ans: Each filter will has 27 + 1 (bias) = 28 parameters. Since this layer has 10 filter so total #
of parameters is 28*10=280.
Q2:
Layer
Output
Volume
Total
Parameters
(with bias)
Total
Parameters
(without bias)
Name
# of
Filters
Filter
size
Stride
Padding
Input
NA
32x32x1
NA
NA
Conv1
6
5x5x1
1
0
28x28x6
(5x5x1+1)x6
(5x5x1)x6=150
Pool1
NA
2x2
2
0
14x14x6
0
0
Conv2
16
5x5x6
1
0
10x10x16
(5x5x6+1)x16
Pool2
NA
2x2
2
0
5x5x16
0
0
FC3
NA
NA
NA
NA
120
48120
48000
FC4
NA
NA
NA
NA
84
(120x84)+84
120x84
