Joone vs. Encog - production apps

Question

Has anyone used both Joone and Encog in production applications? Is there something you needed to do in a production app with one or the other where the product fell short?

Answer 1

JOONE seems to be mostly dead, its been on "release candidate 1"(with lots of bugs) since Januarl 2007. Encog just released 2.0-beta2, and is an actively supported product. I would go with Encog. http://www.encog.org

Answer 2

This article might be helpful as well.

http://www.dzone.com/links/choosing_an_open_source_neural_network_framework.html

Answer 3

Looked at both. Joone is pretty dormant though has some nice design features in terms of how inputs are managed and normalized. Encog is active and has promise. It's various levels of abstraction are well suited to overriding for your own purposes though you may find yourself spending a lot of time figuring out the various layers of abstracted classes before you get started. Additionally, because of this layering I'm not sure performance will be spectacular though that can be overcome by rewriting the appropriate classes for your specific app.

Answer 4

There is an article that compares all 3 available neural network frameworks: http://www.codeproject.com/KB/recipes/xor-encog-neuroph-joone.aspx