There is the watchmaker package.  It seems to be a moderately complete
package for traditional Holland style genetic algorithms but has not seen a
lot of interest in the community and is probably not a super screaming fast
implementation.

As far as neural networks are concerned, the guys who wrote Neuroph are
talking about merging with Mahout.  See
https://issues.apache.org/jira/browse/MAHOUT-383

Somewhat related, we have a stochastic gradient descent implementation of
logistic regression that is about to go out.  You can view that as a single
layer neural network and it should be reasonably easy to extend that code to
full-scale neural networks.

On Tue, Aug 3, 2010 at 8:08 PM, Saikat Kanjilal <[email protected]> wrote:

> In reviewing the list of algorithms currently in mahout I didn't come
> across neural network implementations or genetic algorithms, I was wondering
> if anyone in the community is thinking about implementations in these areas
> as related to solving problems in image processing related to Image
> similarity or greedy algorithms that converge toward a singular solution.

Reply via email to