back to article Do the numbers, Einstein: AI is more than maths as some know it

Microsoft arrived on the graph-database scene last month. Three years in the making, Microsoft released Trinity under a typical-by-now-of-Microsoft-boring-trade name of Graph Engine. Already on that scene are Neo4J, MarkLogic, Oracle, SAP and Teradata - among others. Driving Microsoft, like those before, is the desire to …

  1. Dave 126 Silver badge

    A good article.

    In fact, it directly answers a question a Reg reader posted a a few weeks back in response to the article https://www.theregister.co.uk/2017/02/17/machine_learning_graphical_eye_candy/ (The article featured images of graphs by Graphcore, but so complex they resemble slices of brain more than they do a road network)

    The reader asked "What does 'graph' mean in this context?" and this article answers it well.

    Images are important. I remember as a child and seeing images that represented topology - a teacup being topologically similar to a doughnut, a teapot being similar to a figure-of-eight pretzel. Wow, I thought, this is maths? Cool! I thought maths was just boring numbers!

  2. Hugh Barnard

    Good introduction, but worth saying that graphs are very general and powerful. For example, some tyoes of zero knowledge proof are graph theory based too: http://tinyurl.com/lq22k44

  3. Destroy All Monsters Silver badge
    Windows

    Conceptually, it's all about graphs.

    Nice intro.

    Anyone who still remembers that the absolutely marvelous multifunction transformer truck that is Prolog and the related sub-enginery (Answer Set Programming, Constraint Logic Programming and Constraint Handling Rules, all of which I invite the esteemed readership to get acquainted with at least in passim) exist and is used in applications more interesting than getting rounded corners with graded coloring on-screen won't be too surprised.

    1. Schultz
      Thumb Up

      Re: Conceptually, it's all about graphs.

      One up for mentioning Prolog. But I have mixed memories, the mind bends dangerously when confronted with too much logic.

    2. Anonymous Coward
      Anonymous Coward

      Re: Conceptually, it's all about graphs.

      I love Prolog, and I think that learning it really helps with understanding and coding abstract problems. Another nice thing about it is if you can imagine a solution in Prolog, you can generally code it up using closures in more modern languages. For example, here's a proof-of-concept that I thought up in Prolog and implemented in Perl. It implements an agent (or bot, if you prefer) that traverses a 3-coloured and/or 7-coloured regular hexagonal tiling using a quine-like operator.

  4. elDog Silver badge

    A breath of knowledge/science/maths at El Reg! Bravo!

    I'm fascinated by these newfangled graph engines. We used to have a CODASYL networked database back in the 70s that handled a lot of the cases that Neo4J/etc. does now - not as slickly of course.

    Another problem for the yanks: Which state do you have to start from if you want to form a path that only visits every state once? There are several solutions so it's best to think of a commonality between all the solutions. (https://www.cs.cmu.edu/~bryant/boolean/maps.html).

  5. Tony Haines

    Königsberg's bridges

    I reckon I can solve the Königsberg bridge problem. But the solution does involve leaving the city and hiking around the river's source.

  6. Anonymous Coward
    Anonymous Coward

    The frankly wonderful Cliff Stoll (who blows and sells klein bottles for fun) did a very intuitive video on this for Numberphile, if you're more the visual sort.

    https://www.youtube.com/watch?v=W18FDEA1jRQ

  7. Stevie Silver badge

    Maths isn't about equations you can't understand

    I wish you'd told this to my 6th form calculus teacher at St John Backsides Comprehensive years ago. Could've saved everyone a lot of tedious shouting and threatening and carrying on like a lunatic.

  8. ACZ
    Thumb Up

    Nice intro - thanks

    Nice intro to graph theory and graph databases, thanks. Might go and do some more reading - would love to understand *how* nodes and edges are expressed/stored within graph databases, how the graph database engines work, and how that can then facilitate insight into large complex datasets

    @AC - thanks for the Numberphile video link

    1. Anonymous Coward
      Anonymous Coward

      Re: Nice intro - thanks

      This is a little dated but the fundamentals are still there.

      https://www.slideshare.net/thobe/an-overview-of-neo4j-internals

      The answer is basically linked lists of linked lists of linked lists. It's pointers all the way down. Once you get the data into memory it is *extremely* fast to traverse a graph, but partitioning, sharding and disk IO come with serious compromises, which is where you usually end up comparing the various graph database technologies.

  9. Andy 73

    Not bad

    Not a bad intro, but could have done with a little editing and a couple of illustrations. El. Reg should be able to do a little better than this - the information and understanding is clearly there, but the presentation is a bit behind.

  10. JeffyPoooh Silver badge
    Pint

    A.I. is hard...

    After I make an online purchase of a purple widget, the world's ad broker's A.I. starts showing me dozens and dozens of advertisements for more purple widgets.

    I already have one on the way. Nobody has ever needed more than one purple widget.

    A.I. is hard.

    1. fajensen Silver badge
      Coat

      Re: A.I. is hard...

      A.I. is hard.

      Sure, but, as you now realise: The A.I. doesn't have to work to suck in money from advertisers. In that sense it isn't a hard problem to use "A.I." to support a business model.

      One could, if one was a very sceptical person, imagine that, due to an unfortunate skew in the measurements of when you buy something off Amazon and when those adds show up in your browser, it would appear to be the case that those adds were all shown before your purchase - thus validating the sophistication of the add-pusher "A.I." to the people who splurge money on the ads?

      Bit like "they" do with the "precision attacks" too.

  11. Anonymous Coward
    Anonymous Coward

    Whoa we've solved the TSP?

    Last paragraphs imply that TSP is somehow "solved" or just how it is stored?

    This is a weird article, but other than that mistake I have nothing productive to add about why it is weird.

    AC

    1. Charles 9 Silver badge

      Re: Whoa we've solved the TSP?

      TSP is readily solvable from the general case. The problem with it is that solving for the general case tends to take more time than we have available. TSP isn't unsolvable; it's infeasible for large cases. The complexity of TSP increases on a much-higher-than-linear scale (I think the scale is factorial).

  12. Anonymous Coward
    Anonymous Coward

    "maths" is not "arithmetic"

    'the first thing you have to realise is that "maths" is not "arithmetic" as many might believe'

    I did Maths A and S levels, but those contained a fair bit of arithmetic. It wasn't until I did Maths at uni that I fully realised the truth of the above quote.

    I quickly discovered that the only times I used mental arithmetic in earnest was when shopping or playing darts.

    The desire to maintain my skills in arithmetic justified the purchase of a dart board. :-)

  13. GrumpyOF

    Absolutely...the ability to subtract trip19, trip 17, trip15 and trip13 to derive the best way to get to double16 lives on and will never be forgotten. Bit useless for anything else mind you!

    1. Anonymous Coward
      Anonymous Coward

      "Absolutely...the ability to subtract trip19, trip 17, trip15 and trip13 to derive the best way to get to double16 lives on and will never be forgotten. Bit useless for anything else mind you!"

      You might not think so, but that level of playing with numbers definitely helped when it came to ensuring that batch totals in accounting programs were correct.

  14. Esme

    Nice start to a Friday!

    This last several years I've been feeling I'm getting further and further behind most of you clever (and younger, sob!) IT bods that inhabit the commentardiat here, but this article made me feel reasonably clever for as long as it took to read it, as I understood all that back in senior school (indeed, I taught my physics teacher a thing or two about topology - what? No, in an experiment on magnetism involving very long coils and how opposing magnetic fields interact, get your mind out of the gutter, honestly, some people, tsk!).

    For a few seconds, I basked in the glow of feeling a little smarter than others. Then Destroy All Monsters brought me back to reality. Sanity is restored. I know my place, etc - so back on with me flat cap, and another day's drudgery at the Helldesk coalface.. (to be followed by a healthful wee bottle of wine with dinner, of course!)

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2019