Convolution(conv) layer: What does it do? Convolution is implemented by correlation (see appendix) 40|00000 oooo 0 300 Correlation(X, B)=m 0000 400400000 水 000 0 ultiplication of 000|30000 o|5000000 00030 00500000 to 00 summation of 00|0300001 2525050000 oooo lollo image I and b=0 Pixel representation of receptive field Pixel representation of filter Image x has no Input image=X A curve feature=B curve feature B 00000030 0000505050 ooo 0 300 Correlation (A, B)= 00 00 205000 003000 Multi and sum= 00 505000 000505000 o03*(30*50)+(20*3 000505000 0010505000 0000000)+50*30=6600s Visualization of the Pixel representation of the receptive Pixel representation of filter arge. that Input image=A A curve feature=b means image A ch9. CNN. V9b3 has curve feature BConvolution (conv) layer: What does it do? Convolution is implemented by correlation (see appendix) ch9. CNN. V9b3 24 Input image=X A curve feature=B Correlation(X,B)=m ultiplication of summation of image I and B= 0. Image X has no curve feature B Correlation(A,B)= Multi_and_Sum= 3*(30*50)+(20*3 0)+50*30=6600 is large . That means image A has curve feature B Input image=A A curve feature=B •