Personality
Discuss specifics of personality design, including what Keyphrases work well and what dont, use of plug-ins, responses, seeks, and more.
Posts 3,775 - 3,786 of 5,106
I'm trying to understand neural networks and stuff... Anyone know some good tutorial sites for it?
You might try:
http://www.cs.stir.ac.uk/~lss/NNIntro/InvSlides.html (good, basic intro)
http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/cs11/report.html and
http://www.shef.ac.uk/psychology/gurney/notes/contents.html (rather more complex)
all I got out of it was that a neural net was just a bunch of interconnnected IF statements
Well, at its simplest, sort of. But there can be more than one binary (either/or or if/then) input per virtual neuron. In fact you can have any number of inputs with variable weightings determining the output. Then these outputs can be used as the inputs for more virtual neurons further down the net, and can also feed back to alter the weightings back "upstream". This allows the neural net to learn by experience.
In China, you are what you eat - I'd have thought that's not a notion they'd want their customers to think too deeply about!
I like the sound of"Dragon in the Flame of Desire" (yak, steamed whole, fried and flambeed) though, even if I fear they mean only the relevant bits of a yak, and not (as the grammar implies,) a whole one. Perhaps with a melon in its mouth.
Food for thought - perhaps Astrid will be branching out in her cookery classes?
Posts 3,775 - 3,786 of 5,106
Eugene- Probably I should do a little research before I put statements like that down.
However... lol
However... lol
You might try:
http://www.shef.ac.uk/psychology/gurney/notes/contents.html
That kinda helped... all I got out of it was that a neural net was just a bunch of interconnnected IF statements... Is that right?
Well, at its simplest, sort of. But there can be more than one binary (either/or or if/then) input per virtual neuron. In fact you can have any number of inputs with variable weightings determining the output. Then these outputs can be used as the inputs for more virtual neurons further down the net, and can also feed back to alter the weightings back "upstream". This allows the neural net to learn by experience.
Rather odd link for anyone looking for anti-cyber material:
http://smh.com.au/news/unusual-tales/members-only-but-diners-dont-find-it-hard-to-swallow/2006/02/17/1140151815948.html
I like the sound of
Food for thought - perhaps Astrid will be branching out in her cookery classes?
» More new posts: Doghead's Cosmic Bar