Generally at the end of CNN there is a linear(fully connected layer) layers. But I have recently thought what if i change some convolution layers in CNN with linear layers? Is it useful? if yes where? or if no why not? I would really like to listen your opinion.
Asked
Active
Viewed 25 times
0
-
Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. – Community Jun 14 '22 at 17:04
-
I am sorry "Mr.Bot" You may not understand curiosity but Human can be intrigued by imagination. How can i clarify more? What happened if some convolution layers are replaced by fully connected layers? It can increase the performance? – Seung Hwan Kim Jun 15 '22 at 00:36