I've been trying to find info on how Neural Networks are saposed to work, and so far have found ... nil.
Help, please?
Printable View
I've been trying to find info on how Neural Networks are saposed to work, and so far have found ... nil.
Help, please?
Are you speaking of our own brain Neural networks or machine intelligence?Quote:
Originally Posted by Keeper
both would be nice. Dont they work on similer principles?Quote:
Originally Posted by Howetzer
Well I guess you could say they are trying to work on the same principles.Quote:
Originally Posted by Keeper
Our brain networking, firing neurons, chemicals, dendrites etc are all an elaborate communicating process that receives stores and processes information.
that to is a similar process to the computer and the advancement of AI and "smart computers"/learning computers.
They simulate how the brain functions in many ways. Similar enough it is a bit scary. But computers, as advanced as they have become, even the most powerful computers available today are minimal compared to the complexity, efficiency, and information processing capacity of the human brain.
yes, I remember reading that our brains have over 100 trillion conections, something computers have not even come close to approching.
But how does the conection work? How does the brain store the data, and redirect impulses in a manner that couses thought?
And how are we trying to copy it? What manner have we tryed to immatate them?
Artificial Neural Networks, as their name suggests, is based off of their biological equivalent in creatures
in brief, you have to break all the functions of a system down into it's smallest component parts
each part (called a node, or neuron) does one very specific task
each node takes in a known type of input, does it's specific process, and outputs a known type of output
these nodes are arranged in a web-type of structure
with nodes that accept the same input (but output differently) on the same "level"
any node that outputs what another node can accept as input, is connected to that node
the goal of any neural network, is to find the best path from start to finish
through as few nodes as possible
when ever a successful travel is made from start to finish, the connections between the nodes is strengthened
so the next time the system is given a similar input, it knows what path to try first
this is the learning side of neural networks
the human brain is a lot more complex, but the principle's the same
A baby, when it starts walking, will pull itself up onto it's feet using a peice of furniture
and attempt to walk
it has many different muscles it can use, and it can use each muscle in several different ways.
the goal is to find the best path between all the muscle controlling neurons in it's brain to enable it to walk successfully
once it finds a path, the connections between the neurons will strengthen
Neural networks are pattern finders
you model a complex situation with a web of interconnected neurons
and the system will seemingly understand the situation and come up with an "answer" (read "answer" how you like, it traverses the paths and reaches the end)
Neural networking has stalled somewhat in recent years
due to computing limitations
and it's primarily used for pattern recognition
ideally, the neurons themselves should be able to adapt themselves
based on global criteria
at the moment, neurons are static, and don't change
A common joke is that for a neural network to function properly, you have to tell it the answer before hand
probably the best known public use of neural networks was in the game Creatures
http://en.wikipedia.org/wiki/Creatures_%28...life_program%29
however, if you're interested
there's a derivative of XML called AIML
Artificial Intelligence Markup Language
that's used in Bot programs, like the AliceBot
http://www.alicebot.org/
"How do Neural Nets work?"
They don't. HAW HAW!
hmm....Quote:
Originally Posted by Kaniaz
"I don't know, but if you give me the answer, I can tell you how you got to that answer" may have been funnier
or maybe not
Yeah probably. Hey, I don't work good under pressure. Besides, I am totally not insecure. If someone doesn't like it, hey, that's their problem.
You liked it, right?