Ask Your Question

Convolutional Neural Nets versus kernel-based morphology transformations. What is better?

asked 2018-04-12 00:52:50 -0600

updated 2018-04-26 12:39:48 -0600

If somebody has experience or just tried to compare for himself, can you list the differences? Typical applications, limitations, different output ...

Programmatically, they are the same so I wonder, are there any differences besides the names used?

edit retag flag offensive close merge delete


please add some context.

berak gravatar imageberak ( 2018-04-12 02:31:46 -0600 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2018-04-26 12:40:46 -0600

2 main differences may be outlined. 1.NN can work in 2 modes: learning and production. If convolution is used in learning, this is a completely different story. 2.As to production mode, kernel-based transformations are more general. You can write a sophisticated processing hardly imaginable with NN, but the nets are more natural. If you restrict the solution to this class, probably it will work like human vision.

edit flag offensive delete link more


I never heared of production, but do you mean inference?

StevenPuttemans gravatar imageStevenPuttemans ( 2018-04-27 06:17:28 -0600 )edit

No. During learning you prepare the net, that is write something into its associative memory. Then, you stop training and make use of your product. Production system is a standard term in common symbolic programming. They are used for such tasks as parsing. Neural net in fact does something similar, only instead of rules it uses learned associations.

ya_ocv_user gravatar imageya_ocv_user ( 2018-05-02 05:59:52 -0600 )edit

Question Tools

1 follower


Asked: 2018-04-12 00:52:50 -0600

Seen: 394 times

Last updated: Apr 26 '18