I am doing a Lazarus on the graph question to seriously doubt the utility of graphs as an AGI tool, although it is a very natural narrow AI tool. If anything I think it may be too fuzzy to handle what I think is (one of) the core(s) of human-like intelligence, namely a rigid ontology- unless again you are willing to give up almost everything that's graphy about graphs and just do the ontology with them. But I would like to see more experimental results about the kinds of AGI-roadmap problems where graphs may excel, and again I would be keen to look at general game playing with a view to eventual deployment as game AIs, the latter "game" as in billion-dollar games.
AT > On 10.01.2014, at 20:01, Samantha Atkins <sjatk...@gmail.com> wrote: > > They also seem to have some good proprietary for finding and quickly > projecting out the aspects of a high dimensional model that are the most > important to a particular set of criteria. Which I see would be highly > important to AGI and many other projects. > > >> On Thu, Jan 9, 2014 at 5:30 PM, Samantha Atkins <sjatk...@gmail.com> wrote: >> I don't see any magic at YarcData that fixes the general problem of heavy >> graph traversal of nodes and links that span different machines. YarcData >> as far as I can tell sucks information into big memory stores. Which really >> is a variation imho of the old vertical scalability answer of just buying a >> bigger machine. :) >> >> Am I missing something important? >> >> >> On Wed, Jan 8, 2014 at 12:47 PM, Ben Goertzel <b...@goertzel.org> wrote: >>> >>> The scalability of otherwise of graph knowledge stores depends >>> significantly on the hardware in question... >>> >>> If one uses hardware with massive hardware multithreading, e.g. the Yarcdata >>> >>> http://www.yarcdata.com >>> >>> one will find massive graph databases quite efficient to deal with.... >>> Currently commodity hardware just happens not to be very efficient for >>> handling the common queries and manipulations one wants to do on large >>> graphs... >>> >>> I tend to agree that graphs and hypergraphs are a natural representation >>> for AGI. And I project that commodity hardware with high degrees of >>> hardware multithreading is coming before too too long .. stuff like the >>> Xeon Phi is a small step in that direction... >>> >>> -- Ben G >>> >>> >>>> On Wed, Jan 8, 2014 at 11:33 AM, Samantha Atkins <sjatk...@gmail.com> >>>> wrote: >>>> Best representation of some information is a function of the operational >>>> requirements on that information and subject to change over time. Perhaps >>>> the best that can be achieved is efficient means to transform information >>>> structures to be optimal for primary needs of the moment on the fly. It >>>> is a trade-off of efficiency gained - cost of transformation. Even given >>>> a particular representation there are interesting partitioning and >>>> grouping questions relevant to real world efficiency. With sufficient >>>> metadata and collected usage information it is conceivable information >>>> models can transform to optimize themselves on the fly. Projects I have >>>> known of in the past to do this did not seem highly successful. The >>>> exception is caching in faster access media results previously computed. >>>> >>>> >>>> Graphs are not so scalable. At least this is the word from the graph >>>> database community. >>>> >>>> >>>>> On Mon, Dec 16, 2013 at 3:37 PM, Aaron Hosford <hosfor...@gmail.com> >>>>> wrote: >>>>> It can be represented as such. But propositions are not generally >>>>> restricted to pairs of items. If you have a more complex logical >>>>> structure, you would need more edges, whereas a single, more complex >>>>> proposition would likely be used.* Mathematically, the two >>>>> representational methods are identical in expressive power, but they lend >>>>> themselves to different sorts of operations and have different >>>>> representational advantages. A list can be used to represent a set, and >>>>> sets can be used to represent a list, but if you need to represent lists, >>>>> you probably don't want to go with sets, since you'll end up with >>>>> something really awkward. Likewise, weighted knowledge graphs are >>>>> naturally better suited to the representation of ambiguous information >>>>> than logical propositions. >>>>> >>>>> *(Or do I have my terminology skewed? I used to call vertices nodes and >>>>> edges links, to the confusion of anyone familiar with graph theory, since >>>>> I don't get many opportunities to be corrected on my choice of words with >>>>> regards to this stuff. By proposition, I mean a logical expression made >>>>> up of Boolean operators applied to symbols, and possibly including >>>>> references to truth functions.) >>>>> >>>>> >>>>>> On Mon, Dec 16, 2013 at 5:24 PM, Piaget Modeler >>>>>> <piagetmode...@hotmail.com> wrote: >>>>>> One question, isn't an edge the same thing as a proposition? >>>>>> >>>>>> ~PM >>>>>> >>>>>> >>>>>> Representing this in a propositional form (which is the interpretation I >>>>>> give to the phrase "symbolic representation" -- correct me if I am >>>>>> wrong) would require two separate propositions, each with a graded truth >>>>>> value, but with no clear connection to each other. To record the >>>>>> relationship between them -- that they are both candidate bindings for >>>>>> the anaphoric noun phrase -- a third proposition would have to be >>>>>> created which contained the first two as sub-propositions explicitly >>>>>> related to the noun phrase and their respective weights. But with the >>>>>> weights contained inside the proposition, each time a change was made to >>>>>> those weights, we would have a new proposition, and would have to throw >>>>>> away the old one for being out of date. (The other option being to make >>>>>> propositions mutable, which would make search a nightmare.) >>>>>> Additionally, the number of ways to represent the same information would >>>>>> grow combinatorically with each additional anaphoric binding option, >>>>>> since the relationships are commutative. Trying to connect this >>>>>> information together with other anaphoric ambiguities in the same >>>>>> sentence would add another layer of combinatorics on top of the one we >>>>>> already have. The proposition used to represent the full, ambiguous >>>>>> meaning of a single sentence would be monstrous. >>>>>> >>>>>> Instead, with a graph, I can represent each option with a single >>>>>> weighted edge, and there is precisely one, maximally compact >>>>>> representation no matter how many options we have for any number of >>>>>> anaphoric noun phrases. Making a change requires only a modification of >>>>>> a single weight, or the addition of a single new edge, operating >>>>>> in-place without significant effect on other nearby information >>>>>> structures. As another advantage, when using graph form, we can take >>>>>> advantage of the many algorithms from graph theory, and the explicit >>>>>> locality of reference for related information, greatly speeding up and >>>>>> simplifying searches for relevant information. I can have a vertex to >>>>>> represent alligators, and all the information I know about alligators >>>>>> connected directly to it, meaning that the system only needs to search >>>>>> vertices connected to the alligator vertex for relevant information, >>>>>> rather than all information in the entire database. I can use spreading >>>>>> activation to quickly find all the vertices that relate both to >>>>>> alligators and to steeplechases, making it easy to determine whether >>>>>> alligators can jump or run and thereby participate in such a race. (A >>>>>> physical or other special-purpose model of alligator behavior could have >>>>>> a reference to it stored under a particular vertex, making that >>>>>> accessible just as quickly as any other data about them.) And if I learn >>>>>> something new about alligators, I can add it without interfering with >>>>>> the other information already present and immediately know which >>>>>> information needs to be collated with the new data. Finally, it should >>>>>> be clear that since graphs can be used to represent computer programs >>>>>> (flowcharts) and data structures (pointer indirection networks), graphs >>>>>> are representationally complete -- they can represent any thing or >>>>>> process that can be represented on a computer in any way. So if our >>>>>> understanding can be modeled computationally, as I believe it can, then >>>>>> graphs are up to the job. >>>>>> AGI | Archives | Modify Your Subscription >>>>> >>>>> AGI | Archives | Modify Your Subscription >>>> >>>> AGI | Archives | Modify Your Subscription >>> >>> >>> >>> -- >>> Ben Goertzel, PhD >>> http://goertzel.org >>> >>> "In an insane world, the sane man must appear to be insane". -- Capt. James >>> T. Kirk >>> 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
