This is quite a broad topic, and it helps to have both some theoretical understanding of the mechanisms and a reasonably good idea of what it is you are trying to achieve (and some patience with the neural net learning process). That said, Brian Schott has put together an example and a lab (which I see he has mentioned while I was composing this message).
That said, there are other options which might or might not be better for whatever it is you are specifically trying to do (including genetic searches - not to be confused with biological genetics, linear algebra, hill climbing and so on...). https://en.wikipedia.org/wiki/Genetic_algorithm http://www.wikihow.com/Divide-Matrices (but note that the first sentence there is false. J for example represents matrix division using %. and the process is faster than finding the matrix inverse). https://en.wikipedia.org/wiki/Hill_climbing That said... if you can pose a specific example of what you are trying to do, (or specific examples, if we seem to be missing the point), we might be able to recommend some specific code. I hope this helps, -- Raul On Mon, Jul 24, 2017 at 5:00 PM, Brian Babiak <bdbab...@gmail.com> wrote: > Can anyone direct me to the possibility of neural nets/supervised learning > in j? The problem I'm trying to solve is there is an input table/matrix and > a value assigned to that matrix. I would like to do supervised learning on > a neural net to optimize the difference to the known value. Any examples or > information about neural nets and supervised learning in j would be > appreciated! > > > Brian Babiak MD > https://drbabiak.com/ <http://drbabiak.com> > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
