Second: what is the effect of kernel size on feature

Question

Machine Learning

D

Second: what is the effect of kernel size on feature

determination?

0

19.03.2020

1 ответов

14 просмотров

David · Accepted Answer

The first layer captures the most basic elements, edges, lines, etc, so you don't need, in theory, that many filters. As you move deeper, the CNN learns features more abstract, eyes, nose, for instance, such things, and when you go deeper, then the filters captures concepts such persons, cars, etc.There are many more features to detect in the deeper layers than at the first layer, that's the reason you will see usually increasing niumbre of filters as you go deep.

Похожие чаты

Second: what is the effect of kernel size on feature

1 ответов

Похожие вопросы