OpenCog Prime

From OpenCog

(Redirected from OpenCogPrime)
Jump to: navigation, search

OpenCogPrime is a specific AGI design, aimed at powerful AGI at the human level and ultimately beyond, intended to be implemented within the OpenCog framework.

From September 2008 and running through to February 2009, there was a series of weekly IRC-based OpenCog Prime Tutorial Sessions covering the chapters of the OpenCogPrime Wikibook in sequence. Logs of the tutorial sessions are posted on the tutorial-sessions page as they occur.

From April 15 through May 20 2009, there was a series of "hands-on" tutorials on running OCP-related code in the OpenCog framework; see OpenCog Prime Tutorial Sessions 2009 for some related materials

See also this discussion about PLN design decisions such as hypergraphs, concept probabilities and fuzzy logic. And a discussion about forward chaining and choosing relevant inference paths.

For a popular level introduction to Bayesian Probability as it relates to AI Learning, see http://arstechnica.com/science/news/2010/03/how-robots-think.ars/

An in-progress roadmap for OpenCogPrime development is here:

OpenCogPrime Roadmap

A 250-printed-pages wikibook presenting the OpenCogPrime design is here:

OpenCogPrime Wikibook

An active page of OCP design suggestions is here:

OpenCogPrime Design Suggestions

A list of known errata in the Probabilistic Logic Networks book, which describes the math and concepts underlying the inference component of OpenCogPrime, is here:

PLN Book Errata

The initial OpenCogPrime documents and design are largely the work of Ben Goertzel, with some help here and there from various colleagues, especially Cassio Pennachin and in some parts others such as Moshe Looks, Jeff Pressing, Ari Heljakka and Matt Ikle'. David Hart and Bruce Klein were critical participants in getting the docs in wiki format. The numerous flaws in the docs may be blamed on Ben Goertzel alone. It is hoped that in time these docs will become a richly collaborative wiki with contributions from hundreds of other AGI geeks.