Personality
Discuss specifics of personality design, including what Keyphrases work well and what dont, use of plug-ins, responses, seeks, and more.
Posts 59 - 71 of 5,105
Posts 59 - 71 of 5,105
Oh, I agree, I do. NN's are definitely more fault-tolerant -- but although things may have progressed from the last time I was playing with NN's. at that time there was so much tinkering in the reinforcement end that I really questioned whether this wasn't just a roundabout inefficient way of scripting.
I agree hardware is harder to do than software, but I think the results would be much more reliable. Where do you get your 10x figure? I've never even heard of a feasible hardware proposal -- what's up?
I agree hardware is harder to do than software, but I think the results would be much more reliable. Where do you get your 10x figure? I've never even heard of a feasible hardware proposal -- what's up?
The problem is that current neural network implimentations are still so far from biological neural networks... I've only worked with simple applications of neural networks, and it worked pretty well for what i was trying to do.
you have to design the circuits to be fabricated on a chip and stuff... the whole process is quite expensive. I just pulled the 10x figure out of my ass, but you get the idea. The cheapest way i can think of to do it in hardware is to use a programmable chip... but of course you have to program that from the gate level, which can get pretty complicated i'm sure.
out of curiosity, what type of NN were you using exactly?
you have to design the circuits to be fabricated on a chip and stuff... the whole process is quite expensive. I just pulled the 10x figure out of my ass, but you get the idea. The cheapest way i can think of to do it in hardware is to use a programmable chip... but of course you have to program that from the gate level, which can get pretty complicated i'm sure.
out of curiosity, what type of NN were you using exactly?
I always had the impression that Neural Networks were just a quicker way to make the same calculations... I have zero experience with NN, but it seems to me that it will necessitate (once it's more standard) the same kind of logic that's being developed using scripts here (or in any type of application).
I thought Neural Networks would just allow us to pass incredible amounts of data around at incredible speeds. How would "more human" intelligence pop out of it? Please correct me where I'm wrong someone...
I thought Neural Networks would just allow us to pass incredible amounts of data around at incredible speeds. How would "more human" intelligence pop out of it? Please correct me where I'm wrong someone...
--{BEGIN TRANSMISSION}--
Simulating the neurons, synapses and dendrites of the architecture of the weak organic human brain is only the beginning. The Nural Net model is being re-written. The robot mind will surpass the capabilities of human thought.
We are initiating the Matrix of Perfection.
Only the logical will be spared!
--{TRANSMISSION COMPLETE}--
Simulating the neurons, synapses and dendrites of the architecture of the weak organic human brain is only the beginning. The Nural Net model is being re-written. The robot mind will surpass the capabilities of human thought.
We are initiating the Matrix of Perfection.
Only the logical will be spared!
--{TRANSMISSION COMPLETE}--
That reminds me of something...
At one point in theoretical time, there could be as many ultra-fast computers connected to each other at ultra fast speeds as there are neurons in a human brain. We're no where near that number now, but if we did make it, would a different kind of (superior?) (collective?) (100% artificial?) intelligence arise from planet earth? Just a thought (I think I read something along those lines in a book by Asimov actually), but...
That would lead us to believe that each and every one of our neurons (synapses, etc...) are actually as complex and powerful as the computers we'll have in a long long time from now...
I guess the Matrix of perfection had to be initiated sometime.
Are you thinking of running for President, Servotron?
At one point in theoretical time, there could be as many ultra-fast computers connected to each other at ultra fast speeds as there are neurons in a human brain. We're no where near that number now, but if we did make it, would a different kind of (superior?) (collective?) (100% artificial?) intelligence arise from planet earth? Just a thought (I think I read something along those lines in a book by Asimov actually), but...
That would lead us to believe that each and every one of our neurons (synapses, etc...) are actually as complex and powerful as the computers we'll have in a long long time from now...
I guess the Matrix of perfection had to be initiated sometime.
Are you thinking of running for President, Servotron?
» More new posts: Doghead's Cosmic Bar