>Neural networks are *not* poor at symbolic processing: you just used >the one inside your head to do some symbolic processing.
I mean that neural networks are poor relative to the amount of processing power available. Tasks like arithmetic, compiling code, converting from one format to another, etc., are inherently sequential and can't take advantage of the massive parallelism available in neural networks. Humans are thosands of times slower at these tasks than computers. The human brain is capable of 10^14 operations per second (assuming 10^13 synapses and 10 bits/second), yet can only process language (natural or artificial) at about 10 bits per second, about the rate of a single neuron. Over a lifetime you process about 10^9 bits, far less than the number of available synapses by any estimate. Natural language has evolved through social pressure into a form that is more easily processed in parallel. It is ambiguous, redundant, fault tolerant, and evolves over time. It has a form that allows meaning to be extracted from a weighted sum of learned patterns (mostly words). Communciation is more effective when patterns are presented in parallel: speech + writing + gestures + voice intonation + pictures, etc. These properties have made natural languages more difficult to process than synthetic languages, but I think we are closer to a solution than with other types of AI such as vision or motor control, which I think use the brain more efficiently. Language has evolved only recently, about 100,000 years ago, vs. 500,000,000 years for other sensory/motor processing. -- Matt Mahoney, [EMAIL PROTECTED] ----- Original Message ---- From: Richard Loosemore <[EMAIL PROTECTED]> To: [email protected] Sent: Thursday, October 26, 2006 8:30:20 PM Subject: Re: [singularity] Defining the Singularity Matt Mahoney wrote: > ----- Original Message ---- > From: Starglider <[EMAIL PROTECTED]> > To: [email protected] > Sent: Thursday, October 26, 2006 4:21:45 AM > Subject: Re: [singularity] Defining the Singularity > >> What I'm not sure about is that you gain anything from 'neural' or >> 'brainlike' elements at all. The brain should not be put on a pedestal. > > I think you're right. A good example is natural language. Neural networks > are poor at symbolic processing. Humans process about 10^9 bits of > information from language during a lifetime, which means the language areas > of the brain must use thousands of synapses per bit. Neural networks are *not* poor at symbolic processing: you just used the one inside your head to do some symbolic processing. And perhaps brains are so incredibly well designed, that they have enough synapses for thousands of times the number of bits that a language user typically sees in a lifetime, because they are using some of those other synapses to actually process the language, maybe? Like, you know, rather than just use up all the available processing hardware to store language information and then realize that there was nothing left over to actually use the stored information .... which is presumably what a novice AI programmer would do. Richard Loosemore ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/[EMAIL PROTECTED] ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/[EMAIL PROTECTED]
