Maybe parts of a chair are a bad place to start. Maybe you have to look at the whole thing, Mike. Contextual information tells you what a part of an object is, not the part by itself. Is that vertical piece of wood the leg of a chair or a baseball bat leaned against something? You'll never know until you look at the scene it appears in. Is "go" a noun or a verb? You'll never know until you put it into a sentence. Have a "go" at it. It's the relationships between the elements which is key. That's where the pattern lies. Not in any one element looking a certain way.
You keep taking things out of their natural contexts, and then demanding we show how our approaches would recognize them. My parser recognizes "go" as both a noun and a verb, depending on use. It builds up a set of trees that show how the elements, the words, must be connected into bigger elements, the phrases, and finally on to the entirety, the sentence. At each step, it weeds out the combinations that make no sense. What we need is not the ability to recognize the tops of chairs taken out of context. (I wouldn't recognize those blobs if I didn't know what they are already, so how is software that's meant to copy my abilities going to do it?) What we need is a "parser" for images. Something that identifies the outlines and fields that make up an image as the base-level elements, and builds up higher level elements out of them, excluding combinations that make no sense along the way, until the whole image has been explained in terms of the *relationships* between those elements. Only then we can look at how the parts fit together to determine an object's shape, and from its shape, its characteristics (being self supporting, being artifactual, having "sittability", etc.), and from its shape and/or characteristics, its identity. On Tue, Oct 23, 2012 at 7:01 PM, Mike Tintner <[email protected]>wrote: > PM: “Does that answer your question, if not why not? These are the > elements...” > > [PM, I’ll come back to this later on] > > > Let me try to pin down the fundamental misunderstanding by focussing on > a fact that is all too painfully true: > > 1. THE ELEMENTS OF THE WORLD: why has every would-be AGI system always > ended up as being some version of a “blocks world”, going back to Winograd? > Look at Ben’s demos. Blocks worlds. Look at some current robotics projects > and papers – blocks worlds. Many AGI-er novices also start testing their > ideas on blocks worlds, or blocks scenes. Remember Dave? Sergio’s test > scene for his theory was also essentially a blocks world, even if the > blocks were dots. AGI-ers themselves recoil from these results – realise > there is something fundamentally too simple about them. > > The main reason AGI-ers invariably end up dealing with blocks worlds, is > because all their systems, logical, maths, geometry, and computing, > presuppose a world made of uniform elements – uniform blocks. – uniform > numerical, logical and formal/geometrical units. And this permeates the > theorising of AGI-ers at every level. They have blocks hammers so they > presuppose the world is made of blocks nails. > > When B & B, AND kurzweil AND Hawkins start talking about a world made of > patterns which is analysed by a pattern-finding brain, they are > presupposing a blocks world. Patterns are made of uniform blocks. > > And when Jim & co start talking of a world that can be conceptualised > into semantic networks, they are also presupposing a blocks world, which > can be analysed in terms of uniform semantic blocks. “Trees” that can be > analysed into uniform “trunks” and “branches” and “leaves.” “Chairs” that > can be analysed into uniform “legs” and “seats” and “arms” and “backs.” > > Note that the blocks can be very sophisticated – you can have > complicated fractal blocks, and intricately curved blocks, as well as Lego > blocks. > > Nevertheless the assumption is always that the world is made up of > uniform blocks – and it doesn’t matter whether it’s real world scenes, or > texts of words, or conversations, sequences of sounds, or complex movements > – the assumption is that they can be analysed into uniform, common blocks. > > 2. THE FAILURE OF THE UNIFORM BLOCKS APPROACH. But wherever you look > this approach is failing and has always failed - the uniform block > approach isn’t working in visual object recognition, or text/language > understanding, or conversation, or creativity, or metaphors.. > > 3.AGI-ERS AREN’T LOOKING AT THE REAL WORLD/ REAL OBJECTS. The first > problem I am having is just getting AGI-ers to **look** at the world – to > look at the scenes of the real world, and the physical objects that > comprise them. > > We have just seen a classic example in PM’s response – I clearly and at > length twice asked him to show how his method could identify elements of > the CHAIRS given. > > He has just replied at length about how his system analyses forms into > elements – but NOT elements of the chairs – just purely general elements. > He has **totally failed to look at the chairs**, totally failed to deal > with the problem. And **he is genuinely unaware that he hasn’t looked at > the chairs, isn’t dealing with them at all.** I take him as being genuine – > and he has totally ignored the problem. > > What is happening is that over and over people are looking not at the > real objects given – but instead thinking “how can my method/ my approach > apply to these objects?” They look at their APPROACH, not at the real > objects, at their hammers, not the actual nails, just as PM has done. > Their approach is a blocks approach, so it surely must work – they don’t > even need to look at the real objects and see if it really does work. > > 4. CARVE THE WORLD/OBJECTS INTO ELEMENTS. What AGI-ers must do is look > directly at the objects given, and start carving them physically, visually > into elements.. > Let’s simplify and take just the *tops* of the chairs. (N.b. I > appreciate it’s a major problem just to decide what *are* the tops, or how > you should section them into lines, dots, pixels etc. – but these > difficulties will apply no matter how you decide to section these chairs or > other objects] > > Here is an illustration of some of the tops of these chairs > > > [image: Untitled] > > What we have here is clearly not uniform elements or blocks but > fundamentally different elements/tops – fundamentally MULTIFORM ELEMENTS – > and v. definitely non-uniform elements. (And this analysis can be repeated > for every other part of these chairs). > > There is and can be therefore NO PATTERN to these tops, or to the chairs > as a whole. No common elements, no pattern. > > If you start carving/dissecting the objects and scenes of the real > world directly - from natural landscapes to urban landscapes – you will > similarly find no uniformity, no patterns (except in odd, isolated parts).. > > I have given endless examples – I am willing to give them again. > > But if people refuse to look, refuse to dissect – if people like Ben > refuse, for example, as he did, to look at and analyse patchworks > [configurations of MULTIFORM elements] and just blindly insist that they > are uniform, patterned, without discussion/analysis – then I am at a loss. > And his refusal is typical. > > But the reality is that we do not live in a uniform blocks/bricks > worlds, we live in a multiform, blobby/ rocks world. Just do a little > carving – leaving your chosen “method/approach” totally to one side - and > that will be self-evident. > > And the task of AGI is to deal with that problem wh. has defied all > attempts at uniform conceptualisation/object recognition and creativity – > deal with the problem of a multiform world – not turn a blind Nelsonian eye > to it. > > > > > > > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/23050605-bcb45fb4> | > Modify<https://www.listbox.com/member/?&;>Your Subscription > <http://www.listbox.com> > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
<<Untitled[1].jpg>>
