It was exactly an hour since I started the program and the timeout is set to 60 mins

Regards,
Xander

Is there a way to speed this stuff up? It's doing 100 counts per 10 mins. At this rate it will take 100 mins to do all of them. Surely it doesn't take that long to recalc the whole net? I don't really care about the displaying stuff.

Regards,
Xander

try the newer version in the neural net thread. Better performance, you'll get the run time errors quicker

the delay time is calculated sharper and even may be adjusted by that global variable.

Can you give me the link to that thread, please?

Edit: It would've been nice if you had told me up front that there was a better version to test rather than have me spent two hours waiting for this thing to complete properly. On second thoughts, just mail me the program you want tested, not going to dig around the forums copying and pasting stuff into an editor window. You have my address

the version actually is not better, the teach-in-layout and the menu structure are a little different.

the calculations are the same, though the calculation time may be set shorter.
As I said before: the same errors, but you get them maybe quicker
But no severe changes.

so don't worry, nothing was in vain

Edit: for mailing, i have to c&p just like you, I'm not sure if I'll find quite the same version on my PC. So plz try c&p, will you? ^^

I'll run it tomorrow, it'll take too long now and I'm going to bed shortly

I modified your program a little more and ripped out some more wait statements that didn't seem to make sense to me.

Regards,
Xander

don't make the WAITs too short.
it's multitasking calculations, you'll have to be sure that 1 calculation is through before the next one starts.

I've run this 5 times to completion now (from 999 to 0) and no error occurs, no unexpected shutdown. In other words, I cannot reproduce this with the current alpha build.

Regards,
Xander

These are good news^^

EDIT: but actually the program should regularly stop before reaching 999 training cycles: in the moment it has learnt the rules by backpropagation...