Epstein Secrets

N nodes are activated and exported to the OpenCog Atomspace • Recognition that a block is present is carried out, either by - PLN inference within OpenCog, drawing the conclusion that a block is present from the exported predicates, using ImplicationLinks comprising a working definition of a "block"

ning to bridge DeSTIN and PLN Ls intended to fit. This architecture is not yet fully implemented, but is a straightforward extension of the current OpenCog architecture for processing data from game worlds IGEA08], and is scheduled for implementation later in 2013 in the course of a funded project invo

ve found is to carefully choose the size penalty function so that the search will tend to restart on simpler programs even if they do not exhibit • Bayes rule as used here is P(MID) = P(M)P(DIA" where Al denotes the Model (the program) and D P(D) denotes the data (the behavior to imitate), here P(D)

his nature that can be effective in this context We have not yet done any sophisticated experiments in this regard - our current experiments using OpenCog to control robots involve cruder integration of OpenCog with perceptual and motor subsystems, rather than the tight hybridization described in Chapt

obabilistic evolutionary program learn- Mg is an effective method The probabilistic evolutionary program learning method we have worked with most in OpenCog is MOSES, and significant evidence has been gathered showing it to be dramatically more effective than genetic programming on relevant classes of p

ation that could potentially be done faster if a lot of the probabilistic knowledge from the AtomSpace were summarized in a Bayes Net. Updating the Bayes net structure can be slow, so this is probably not appropriate for knowledge that is rapidly shifting; but knowledge about properties of blocks str

igence (or any other kind of highly powerful general intelligence), the creation of novel ideas will be paramount. New idea creation takes place in OpenCog via a variety of methods - e.g. inside MOSES which creates new program trees, PLN which creates new logical relationships, ECAN which creates new a

ogPrime because CogPrime's knowledge network is loopy. The peculiarities that allow standard Bayer net belief propagation to work in standard loopy Bayes nets, don't hold up in CogPrime, because of the way you have to update probabilities when you're managing a very large network in interaction with