I have a prototype diagram of my Rete structures. If anyone is interested, I'd like feedback to refine some open issues. I took a look at some papers on Collection Oriented Rete systems by Milinde Tambe and Anurag Archarya which seemed to scale up the discrimination network significantly. Despite Rete being old technology, I think it'll be useful for a Proof of Concept since it is "supposedly" well understood and therefore easier to implement. Theoretically. Perhaps an implementation of Persisted Map/Reduce Rete Nets or Persisted Factor Graphs will follow if this try succeeds. Cheers, ~PM.
------------------------------------------------------------------------------------------------------------------------------------------------Confidential - This message is meant solely for the intended recipient. Please do not copy or forward this message without the consent of the sender. If you have received this message in error, please delete the message and notify the sender. From: [email protected] To: [email protected] Subject: RE: [agi] Rete structures and algorithms Date: Tue, 8 Jan 2013 18:58:25 -0800 Thanks, I'll take a look at Rosenbloom's Sigma. For the PAM-P2's scheme memory I have a specific need, which, if you're interested, I can share with you all offline(just send me an e-mail). For the moment, I'll just continue to read. It's times like these when I wish there was an AGI who could teach meon demand anything I wanted to learn. In the meanwhile, I'll just keep doing the traditional research. Cheers all. Michael Miller / PM Date: Tue, 8 Jan 2013 21:18:34 -0500 Subject: Re: [agi] Rete structures and algorithms From: [email protected] To: [email protected] Rete is an old-fashioned message-passing algorithm used in rule engines like DROOLS, and in the SOAR cognitive architecture.... I understood it in detail decades ago but don't feel like refreshing my memory ;p You may want to check out Paul Rosenbloom's use of message-passing in his recent work on the Sigma cognitive architecture... I doubt very much that such message-passing algorithms are scalable to human-level AGI or anywhere near. -- Ben On Tue, Jan 8, 2013 at 7:11 PM, Piaget Modeler <[email protected]> wrote: I've been reading some PhD theses about Rete and I want to bend the structures and algorithms a bit and was hoping for feedback, in particular on which structures I could modify. Was hoping for a conversation with a little more depth. Cheers, ~ PM. Date: Tue, 8 Jan 2013 17:16:31 -0600 From: [email protected] To: [email protected] Subject: Re: [agi] Rete structures and algorithms I've never used retes so I'm just going off of what I read at Wikipedia and DrDobbs. It seems like the Alpha nodes are just checking for individual conditions (e.g. (a<b), or (c = 1)), the connection to the join nodes could be represented as some type of list (linked list, or list of Keys in a database or dictionary) for each alpha node, and the same for the connection from the join nodes to the beta nodes. On 1/8/2013 11:46 AM, Piaget Modeler wrote: Does anyone here understand the Rete algorithm and data structures very well? If so, kindly answer these questions: 1. What is the process by which alpha memories are formed? 2. How exactly are alpha memories linked to join nodes, and join nodes linked to beta memories. Thanks in advance for the simple explanation. I'm sure I'll have follow up questions. ~PM. ------------------------------------------------------------------------------------------------------------------------------------------------ AGI | Archives | Modify Your Subscription ____________________________________________________________ Woman is 53 But Looks 25 Mom reveals 1 simple wrinkle trick that has angered doctors... ConsumerLifestyleMag.com AGI | Archives | Modify Your Subscription -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
