LiveCode Forums.

Has anyone programmed a Neural Network in LC?

I can't find a simple explanation of what is required
although there are a number of examples in other
languages.

Hi Ros,

look at this thread in the use-list from a long time ago.

http://runtime-revolution.278305.n4.nab ... 85533.html

Kind regards
Bernd

Supposedly one can use HyperNext for Neural Networks:

http://www.tigabyte.com/index.html

Thanks for your help bn and 62.

Lots of reading there (and surprisingly) an insinuation
that LC can't manage to do an NN. My guess is that it
can.

What I'm still looking for is a simple explanation of
what the process is. Other than learning another
language (Python) so far no progress.

As HyperNext is a significantly less powerful development system
than LiveCode (and developed by one person), and it can 'do'
neural networks [ http://www.tigabyte.com/hnfiles/neural.html ]
I would be most surprised if LiveCode could not.

However, Malkom, the person behind HyperNext specifically
developed HN to make neural networks from the start,
which HyperCard-MetaCard-RR/LC was not.

Even if you don't actually 'play' with HyperNext, a read
of that webpage I referenced above is well worth the time.

Just for those who come upon this forum link while searching.

https://livecode.com/how-to-build-your- ... tor-brain/

That was a pretty good read by Hanson, even 3 years later. Not just factual, but actually entertaining.

I am thinking about a possible use where I need to check if a situation with many parameters is fullfilled and how far I am off and do actions accordingly. ... well it is just a though ... I will see if I can use it during some experimentation....

mrcoollion wrote: ↑

Mon Aug 26, 2019 7:44 pm

I am thinking about a possible use where I need to check if a situation with many parameters is fullfilled and how far I am off and do actions accordingly. ... well it is just a though ... I will see if I can use it during some experimentation....

Would that need to be a neural net, or would a Bayesian algo suffice?

Hmmm a Naive Bayes algorithm would probably fit the bill better. That is an excellent suggestion..
I will need to dive into that..
Is there any LC example code of this?

PS.. thanks

Happy to help. I did a quick search but was unable to come up with an existing LiveCode libraries for Bayesian algos.

I wonder, if you found a particularly good library for that written in something else like Python, whether we could all pull together to implement those algos in LC.

"Want to play chess?"--WarGames

Close sphere, it was actually "Shall we play a game?"

Thoughts . . .

http://www.tigabyte.com/hnfiles/neural.html

bogs wrote: ↑

Thu Aug 29, 2019 1:10 pm

Close sphere, it was actually "Shall we play a game?"

ha, yes, you caught me, i was a bit in doubt what it was exactly