http://books.nips.cc/papers/files/nips17/2004_0572.pdf
This isn't exactly AGI, but I know that may AGI designs contain algorithms for automatically clustering multi dimensional data points. In this paper they take pretty much standard spectral clustering but add an extra parameters so that each point has its own scaling factor that depends on how close its neighbours are. With this relatively simple change it does some pretty amazing clustering, such as, concentric circles and one example where there are a few tight blobs of points and then one really big spread out cluster that covers the whole space. It gets the intuitively correct answer in each case. Figure 3 has some nice examples of it working. Cheers Shane Find local movie times and trailers on Yahoo! Movies. http://au.movies.yahoo.com ------- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]