Steve, I skimmed the Microsoft article. Your description seemed accurate. It certainly would be valuable.
I also skimmed your website. It is very impressive. I hope when opencog starts you might join it. >From skimming your website, I didn't really have time to get a feeling for what your architectural approach was. It seems you are plugging together a lot of already available pieces, but it wasn't clear what the AGI glue at the center of it would be. By that I mean, I didn't understand what you were going to do to give the system a generalized, context sensitive, learning, perceiving, generalizing, similarity matching, remembering, mental and physical behavior selecting and controlling, reward based credit assigning, system -- one that would be a generalized, autonomous learning and acting system. You probably have one, but it didn't pop out at me. Ed Porter -----Original Message----- From: Stephen Reed [mailto:[EMAIL PROTECTED] Sent: Tuesday, December 18, 2007 11:20 AM To: agi@v2.listbox.com Subject: Re: [agi] Ed, I've experimented with various Java parallel processing frameworks, most recently the fork-join framework (JSR166) that will be included in Java 7. The Sphinx-4 automatic speech recognition that I use employs a mulit-threaded phoneme scorer in order achieve better performance. For the most part, the overhead precludes using Java parallelism for small-grained computation tasks in the programs that I write. However the Microsoft approach is promising according to this article: http://msdn.microsoft.com/msdnmag/issues/07/10/Futures/default.aspx They include a "Parallel for" statement for the .Net platform that I think will be eventually copied by the Java community. It seems to have low overhead, and I suppose that it works by having a hidden thread pool and VM optimization to execute the designated byte-codes in parallel. I have devised a manycore development platform, in advance of the actual availability of these devices, by assembling a small and very cheap Linux cluster. You can read about it at: http://texai.org/blog/2007/09/21/cheap-cluster . -Steve Stephen L. Reed Artificial Intelligence Researcher http://texai.org/blog http://texai.org 3008 Oak Crest Ave. Austin, Texas, USA 78704 512.791.7860 ----- Original Message ---- From: Ed Porter <[EMAIL PROTECTED]> To: agi@v2.listbox.com Sent: Tuesday, December 18, 2007 9:04:56 AM Subject: [agi] Here is the URL of a Kurzweil-linked NYTimes article about Microsoft's commitment to writing code for many-core processors and how they will enable standard PC's and handhelds to do AI tasks such as vision, speech rec, and semantic content understanding. http://www.nytimes.com/2007/12/17/technology/17chip.html?_r=1&oref=slogin&pa gewanted=print Advances in many-core processing is a positive for the development of AGI, although I believe AGI will be much easier to parallelize than most traditional programming. Ed Porter ----- 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/?&; ____________________________________________________________________________ ________ Looking for last minute shopping deals? Find them fast with Yahoo! Search. http://tools.search.yahoo.com/newsearch/category.php?category=shopping ----- 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/?&; ----- 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=8660244&id_secret=77236144-bbebcf
