tensor-fsmn icon indicating copy to clipboard operation
tensor-fsmn copied to clipboard

why not implement it by cnn

Open hanghang2333 opened this issue 5 years ago • 3 comments

I think the fsmn is similarity to the convolution layer, but I can't find anyone use the conv layer, does I wrong?

hanghang2333 avatar Mar 13 '19 03:03 hanghang2333

SFSMN can be implemented using convolution layer, but vFSMN cannot. The operation is similar as convolution, but it is not the same. Conv ops do multiply and reduce sum in to a scalar, but fsmn reduce sum in one dimension

wd929 avatar May 22 '19 09:05 wd929

SFSMN can be implemented using convolution layer, but vFSMN cannot. The operation is similar as convolution, but it is not the same. Conv ops do multiply and reduce sum in to a scalar, but fsmn reduce sum in one dimension

So vFSMN is same as depth-wise conv?

wizardk avatar May 28 '19 08:05 wizardk

SFSMN can be implemented using convolution layer, but vFSMN cannot. The operation is similar as convolution, but it is not the same. Conv ops do multiply and reduce sum in to a scalar, but fsmn reduce sum in one dimension

So vFSMN is same as depth-wise conv?

You are definitely right.

xx205 avatar Jul 30 '20 06:07 xx205