Pointless Information

[Jazzem]

I know this certainly isn't pointless, but my grandmother died today

[Chmera]

Aww... *hugs*

[namida]

My best advice is, when a relative dies, don't be sad that they're dead, be happy they won't ever have to put up with you again.

[Chmera]

Nice one, Sunrise.  [smiley=winktounge.gif]

[drumnbach]

I know this certainly isn't pointless, but my grandmother died today

It warms my heart to see people grieving their grandparents because I was never close to mine. When they died, I wasn't affected at all. I feel I've missed out on something, especially when I see the relationship that my nephews and nieces have with my parents.

I have a 92 year old STEP grandmother though... I wish she'd hurry up and pop her clogs .

[Timballisto]

Not so pointless request:

Does anyone care to help me out on something? &#A0;I'm trying to build a neural network in Qbasic. &#A0;I made a little experimental one like this:


            &#A0; Layer 1 &#A0; &#A0; Layer 2 &#A0; &#A0; Layer 3
            &#A0; --------      --------      --------

I-----W-----N-----W-----N-----W-----N-----W----O
                  -W- &#A0; &#A0;---- &#A0; -W- &#A0; &#A0;----             &#A0;
                       ----            ----
                  -W- &#A0; &#A0;---- &#A0; -W- &#A0; &#A0;----      
I-----W-----N-----W-----N-----W-----N-----W----O

The W's are the weights, and the N's are the nodes (or perceptrons, or artificial neurons, or whatever it is that you might call them), and the O's are outputs and the I's are inputs. &#A0;I wanted to figure out how to use backpropagation to train the network (in this case train it to produce outputs that were simply the inputs doubled) so I could apply it to a much larger net. &#A0;My stepdad and I ran through an explanation of it, but I can't code it into my program, because the explanations I found confuse me, mostly because it involves calc, and I'm in algebra 2 (wonderful), but also because it's not very clear how all the equations play into each other. &#A0;If anyone could explain how to code for backpropagation, that would be great. &#A0;Heck, if anyone could explain the algorithm itself (keep in mind my level of math) that alone would be great.

[tseug]

I'm back.

Uhh... timb I'd have to experiment.... (btw I'm taknig calc)

[namida]

That doesn't sound pointless. Please post it in the "Pointful Information Topic".

[MC Marshy]

That doesn't sound pointless. Please post it in the "Pointful Information Topic".

I can't see a "Pointful Information Topic"

[Chmera]

Maybe Sunrise was being pointless.  [smiley=winktounge.gif]

[Timballisto]

Thanks Tseug.  If you turn up anything, let me know.

[ccexplore]

Not so pointless request:

Does anyone care to help me out on something? &#A0;I'm trying to build a neural network in Qbasic. &#A0;I made a little experimental one like this:

Is this for a school project or something? &#A0;Just curious, I want to know what you are trying to do with the neural network.

I wish I could help but it's been quite a while since the AI class I took in college.

Unfortunately the backprogation does involve calculus if I recall, at least in its theoretical foundation. &#A0;I'll have to look it up on the web again to refresh myself on the details.

This is kinda ambitious for you, wouldn't you say? &#A0;You said you don't even know calculus that well. &#A0;And even if you do, there's still the separate field of knowledge on how to program the computer to numerically calculate with the equations involved. &#A0;Which is why I'm really curious what the ultimate goal and purpose is of this.

[edit: try http://cortex.snowseed.com/neural_networks.htm, it includes some code which I think you might be able to translate pretty easily to QBasic]

[Timballisto]

Actually, this is for school.  X( .  A bit ambitious?  Yep.  Oh well.  I think I can stil do it...it'll take time of course.  Fortunately the due date for this is still a ways off.  The bigger one that I already have set up (minus the training method) is meant to distinguish between three and four sided polygons (I already drew up 1000 of these using another program I made, so I'm good there).  The smaller one I'm using to experiment with, mostly to figure out how to train the system.  Right now it's meant to simply double the input values.  I'll check out the code and see what I can do with it.

[namida]

His goal is to take over teh world.

[Timballisto]

I have a question:  What does the"->" in the code mean?  Everthing else I'm okay with.