tensorflow - What's difference between Summation and Concatenation at neural network like CNN? -
what's difference between summation , concatenation @ neural network cnns?
for example googlenet's inception module used concatenation, resnet's residual learning used summation.
please teach me.
concatenation means concatenate 2 blobs, after concat have bigger blob contains previous blobs in continuous memory. example:
blob1:
1 2 3
blob2:
4 5 6
blob_res:
1 2 3 4 5 6
summation means element-wise summation, blob1 , blob2 must have exact same shape, , resultant blob has same shape elements a1+b1, a2+b2, ai+bi, ... an+bn. example above,
blob_res:
(1+4) 5 (2+5) 7 (3+6) 9
Comments
Post a Comment