Joseph, On 5/20/08, Joseph Gentle <[EMAIL PROTECTED]> wrote: > > On Sat, May 17, 2008 at 1:37 PM, Steve Richfield > <[EMAIL PROTECTED]> wrote: > > Now, decades later, come the present discussions about patterns, > apparently > > advanced along with the same lines of "thought" that was behind that IQ > test > > so many years ago. Pattern recognition without underlying supporting > theory > > is WORTHLESS (or perhaps worse) except perhaps to suggest possible > > underlying theory. Any good AGI would see a limitless list of possible > next > > items from any real-world list. > > > > Take a sequence of alternating musical notes. Does anyone here REALLY > think > > that someone is going to sit in Carnegie Hall and continuously play two > > alternating notes just because he started out that way? Note that Fur > Elise > > does indeed start out that way. No, at some point, the odds of continuing > > with the same two notes drops quite low. A good music composition program > > might theorize the best sounding subsequent notes, and THOSE would be > much > > more likely than continuing with the same mindless repeating pattern. > > > > <snip /> > > > > Steve Richfield > > > There are two interesting points here. > > The first is that (in my opinion) pattern matching must come first. I > agree that understanding the patterns (the /why/) is important; but > seeing (even unjustified) patterns is crucial. The benchmark I like to > consider is an intelligent system being connected to a noisy camera > for the first time. The noise may overwhelm the pattern matching at > first, but over time it should automatically learn to filter out the > noise regardless of whether or not it understands why the noise exists > at all.
However, without some sort of understanding random noise, and that random noise can exist in a picture, there is no motivation to take the un-obvious measures (e.g. averaging over many frames, clipping peaks, etc) to remove it. Then, after filtration, subsequent analysis must understand that the result comes from a signal processing process that may distort details, so that decisions should NOT be made on the basis of small details. The second interesting point is music. It is my belief that one of the > important properties of music is how it tickles our pattern matching > ability. Imagine a sequence of notes fed into an intelligent system > (perhaps, through speakers into a human mind). The intelligent system > automatically tries to find patterns in the notes such that it can > model the notes (and predict subsequent notes). I have also made this same point in the past. There are many excellent compositions where a repeating theme has more and more complications added to it. BTW, my very first computer program (on an old electromechanical Burroughs E-101) composed rock-n-roll music. We can imagine another sequence of actions of the intelligent system > as it parses subsequent notes in the music. These actions can indicate > three things: > - The note has been correctly predicted > - The note has been incorrectly predicted and the system doesn't know > why that note in particular was played > - The note was incorrectly predicted, but it shows a new pattern which > explains previously incorrectly predicted notes. > > It is my opinion that beautiful music makes an intelligent mind > trigger 'just the right amount' of these second two kinds of > modelling. YES. I also suspect that some of the best pieces that are "on the edge" of this, where some people can follow this process but others can't, and/or where you seem driven to hear the piece over and over to "understand" ever more of its nuances. Steve Richfield ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26 Powered by Listbox: http://www.listbox.com
