Gen Goertzel writes:
Sooooo ... it's interesting and important stuff, but I can't escape the conclusion that it's gotten a bit of a media boost because of who Hawkins is ;-)
I agree with the analysis in Ben's email. I find the Hierarchical Temporal Memory to be an interesting conception because as an AGI fan I like it when somebody comes up with a (somewhat) new type of computing unit and tries to see how far he can push it. Bundling an autocorrelating "spatial" categorizer with an autocorrelating sequence categorizer, calling it a "unit" then building a hierarchy of them in an effort to learn high level concepts directly from perception is very cool work. As far as I know, this type of project always breaks down a couple of levels in, never really coming up with a conceptually satisfying distinction between cats and dogs, but maybe his approach will work better than past attempts. I wish him luck and am eager to see what he comes up with. His book is entertaining -- but why does every book about this type of endeavor have to spend so many pages attacking the history of AI as being filled with misguided buffoonery? It seems like every ambitious project has to start with "Why has AI failed so miserably for 50 years?" It's irritating. ----- 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/?member_id=231415&user_secret=fabd7936
