December 2018
Beginner to intermediate
226 pages
7h 59m
English
The final step is called sharpening. As a result of convolutional shift, the weights ,
, will not be sharp, —in other words, because of the shift, weights focused at a single location will be dispersed into other locations. To mitigate this effect, we perform sharpening. We use a new parameter called
, which should be greater than or equal to 1 to perform sharpening, and can be expressed as follows:

Read now
Unlock full access