EFTA00830443

From: Ben Goertzel To: Jeffrey Epstein <jeevacation@gmail.com> Subject: Re: Brainfood... Date: Thu, 31 Mar 2016 09:39:28 +0000 Attachments: OpenCog_Plan_Slides_v4.pptx; Budget_and_Teams.xlsx And a bit more project-planning stuff for our envisioned 3-year OpenCog initiative (courtesy of Cassio) On Tue, Mar 29, 2016 at 11:37 PM, Ben Goertzel czMIII I> wrote: > Hi Jeffrey, > So, here is some food for thought and discussion... > We are working on getting OpenCog to control the Hanson Robotics robot > heads for embodied (or at least "emheadied"!) natural language > dialogue. This should be working in Summer 2016, maybe sooner. We > hope to demonstrate this to some extent at AGI-16 which will be in New > York this July. To give you some flavor of this work I attach our > Hanson Robotics / OpenCog software plan for the next few months (this > is work funded by Hanson Robotics). > In parallel with that, we have been finally getting some examples of > cognitive synergy to work in OpenCog. > This paper of ours from AGI-14 ended with a bit of a whimper > http://agi-conf.org/2014/wp-contentJuploads/2014/08/harrigan-guiding-agi14.pdf > however we have recently actually gotten this to work. I.e. we have > gotten ECAN (neural net type activation spreading) to successfully > prune the PLN inference tree. If all goes well we will submit a paper > on this to AGI-16 next week. This shows attention allocation and > inference working together. Yay! > This paper of ours from AGI-15 shows PLN being used to estimate the > truth value of a pattern recognized in data by MOSES (our evolutionary > program learning algorithm) > http://agi-conforg/2015/wp-contentJuploads/2015/07/agi15_goertzel_speculative.pdf > It takes about 24 hours to run in the current OpenCog version. > Obviously this is not OK but ifs a start. > Next step is to put these two things together, and use ECAN to help > PLN inference pruning on nontrivial examples like the one from the > above AGI-15 paper. > The attached PPT on some of our biology work contains two parts > -- first part is "straightforward" Al-wise but has quite interesting EFTA00830443 > results biology-wise: i.e. we used MOSES to figure out genes that > distinguish supercentenarians from healthy 80 year olds > second part is more interesting AI-wise ... using PLN to generalize > from MOSES models > Between these papers we see the first concrete examples of "cognitive > synergy" in OpenCog. They are a bit "toy" but they are also real and > actually work in the codebase now ;) > Finally, the other PPT attached shows the plan that Jim Rutt and > Cassio and I are aiming to carry out over the next 3 years. I am > not sure if you intersected with Jim at SFI or not... > http://www.santafe.edu/about/people/profile/Jim%20Rutt > We are aiming to raise $6M for 3 years of work, but I imagine we could > do it for $4M over 3 years if we made copious use of inexpensive > overseas folks. Jim will put in at least $500K of the funds himself > this foundation gives out $500K/year, and he is willing to put at > least 1/3 of that into OpenCog, maybe more). > The goal of this plan is to make OpenCog-based embodied NLP dialogue > work extremely well and hold a conversation demonstrating obvious > understanding of itself and its environment and situation. At the > core this will require getting cognitive synergy between ECAN, MOSES > and PLN to work in the context of visual/auditory perception and NLP > dialogue. This will be both a huge AGI advance and an amazing demo > that will open pretty much any door we want. > Jim Rutt was a good find, not only because he has been donating a bit > of $$ (about $150K last year, none to me personally but he has funded > Nil Geisweiller and Cassio), but because he bring a management and > organization and business sensibility/experience that complements me > well. And he knows enough science and CS and has put in enough time > to thoroughly understand the OpenCog/CogPrime design... > thx! > Ben > -- > Ben Goertzel, PhD > http://goertzel.org Ben Goertzel, PhD http://goertzel.org "I am Ubik. Before the universe was, I am. I made the suns. I made the worlds. I created the lives and the places they inhabit; I move them here, I put them there. They go as I say, then do as I tell them. I am EFTA00830444 the word and my name is never spoken, the name which no one knows. I am called Ubik, but that is not my name. I am. I shall always be." -- Ubik EFTA00830445