Fundations of Convolutional Neural Networks1. Valid Convolution有效卷积input: n×n{n\times n}n×nkernel: f×f{f\times f}f×f (must be odd !)output:{}
Week1: Fundations of Convolutional Neural Networks
1. Convolution on 2D images
- input: n × n {n\times n} n×n
- padding填充: p {p} p
- stride步长: s {s} s
- kernel: f × f {f\times f} f×f (must be odd !)
- output: ⌊ ( n + 2 p − f s + 1 ) ⌋ × ⌊ ( n + 2 p − f s + 1 ) ⌋ {\left \lfloor(\dfrac{n+2p-f}{s}+1)\right\rfloor \times \left \lfloor(\dfrac{n+2p-f}{s}+1)\right\rfloor} ⌊(sn+2p−f+1)⌋×⌊(sn+2p−f+1)⌋(向下取整)
1.1 valid convolution有效卷积
- input: n × n {n\times n} n×n
- padding填充: p = 0 {p=0} p=0
- stride步长: s = 1 {s=1} s=1
- kernel: f × f {f\times f} f×f (must be odd !)
- output: ( n − f + 1 ) × ( n − f + 1 ) {(n-f+1) \times (n-f+1) } (n−f+1)×(n−f+1)
1.2 same convolution相同卷积
- input: n × n {n\times n} n×n
- padding填充: p = ( f − 1 ) / 2 {p=(f-1)/2} p=(f−1)/2
- stride步长: s = 1 {s=1} s=1
- kernel: f × f {f\times f} f×f (must be odd !)
- output: ( n + 2 p − f + 1 ) × ( n + 2 p − f + 1 ) = n × n {(n+2p-f+1) \times (n+2p-f+1) = n\times n} (n+2p−f+1)×(n+2p−f+1)=n×n
2. Convolution vs.cross-correlation卷积vs互相关函数
-
目前我们理解的卷积，其实叫做互相关函数(cross-correlation)，其公式描述如下:
S (
最低0.47元/天 解锁文章
189
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
