CNNConv2DOn this pageConvolution 2D zi,jl=∑m∑nai+m,j+nl−1(ω′)m,nl+bi,jlz^l_{i,j} = \sum_m \sum_n{a^{l-1}_{i+m,j+n} (\omega')^l_{m,n} + b^l_{i,j}}zi,jl=m∑n∑ai+m,j+nl−1(ω′)m,nl+bi,jl Al=σ(Zl)=σ(Al−1∗Wl+Bl)A^l = \sigma (Z^l) = \sigma ( A^{l-1} * W^l + B^l )Al=σ(Zl)=σ(Al−1∗Wl+Bl) Convolution K,K′∈RM×NK(m,n)=K′(M−1−m,N−1−n)K, K' \isin \R^{M \times N} \quad K(m, n) = K'(M-1-m, N-1-n)K,K′∈RM×NK(m,n)=K′(M−1−m,N−1−n) (I∗K)ij=∑m∑nI(i+m,j+n)K(M−1−m,N−1−n)=∑m∑nI(i+m,j+n)K′(m,n)=(I⊗K′)ij\begin{aligned} ( I * K )_{ij} &= \sum_m \sum_n {I(i+m, j+n)K(M-1-m, N-1-n)} \\ &= \sum_m \sum_n {I(i+m, j+n)K'(m, n)} &= (I \otimes K')_{ij} \end{aligned}(I∗K)ij=m∑n∑I(i+m,j+n)K(M−1−m,N−1−n)=m∑n∑I(i+m,j+n)K′(m,n)=(I⊗K′)ij infoCNN 설명에서, 대부분의 그림은 cross-correlation K′K'K′