The thing which typifies the category "chair" is not the shape, but the combination of capability to support a behind and the potential inclination of a person to take advantage of that capability (or intention of the creator to provide such an artifact). These are things that are easy to represent in semantic nets, and difficult to represent as rules about shape.
If I have a representation of an object as a semantic net describing its parts and their physical relationships to each other, I can write a straight forward algorithm to analyze the transitive "supports" and "is connected to" relations in that description to determine whether the spot I intend to sit is supported. I can also determine whether or not my behind, when placed there, will itself be supported, or whether I'll slide off or topple over. The network generating algorithm can be designed to provide the information needed to perform this simulation (simulation being the reason you say images are necessary in the first place). Once the simulation has been performed the first time, the node representing the chair as a whole object can be labeled with a summary of the results, acting as a cache for relevant information so that the expensive operation of full physical simulation can be avoided next time the information is needed. It is this caching ability that gives hierarchical semantic nets their leg up over other ways of representing the problem. -- Sent from my Palm Pre On Oct 23, 2012 11:30 AM, Mike Tintner <[email protected]> wrote: PM & Aaron, You do realise that whatever semantic net system you use must apply to not just one chair, but chair after chair – image after image? Bearing that in mind, explain the elements of your semantic net which you will use to analyse these fairly simple figures as **chairs**:: http://image.shutterstock.com/display_pic_with_logo/95781/95781,1218564477,2/stock-vector-modern-chair-vector-16059484.jpg Let’s label these chairs 1-25 (going L to R from the top down, row after row) Start with just 1. and 2. top left and explain how your net will recognize 2 as another example of 1. How IOW do you define a “chair” in terms of simple abstract forms? Then we can apply your system, successively, to 3. 4. etc. This is the problem that has defeated all AGI-ers and all psychologists and philosophers so far. But Aaron (and PM?) has a semantic net solution to it - if you can solve jungle scenes, this should be a piece of cake. I am saying, Aaron, you do not understand this problem – the problem of visual object recognition/conceptualisation//applicability of semantic nets. You are saying you do – and it’s me who is confused. Show me. From: Piaget Modeler Sent: Tuesday, October 23, 2012 4:41 PM To: AGI Subject: RE: [agi] Re: Superficiality Produces Misunderstanding - Not Good Enough Mike, When you type "Chair" what should happen is the AGI's model should activate the chair concept first at a perceptual level to form the pixels into the words, then at a linguistic level to form letters into a word, then at a conceptual level, then at a simulation level where images of chair instances are evoked. This is just simple activation. Semantic networks tied into perception and simulation would achieve the necessary effect you seek. Transformations on these perception-simulation-semantic networks is what much of Piaget's work was about. ~PM. From: [email protected] To: [email protected] Subject: Re: [agi] Re: Superficiality Produces Misunderstanding - Not Good Enough Date: Tue, 23 Oct 2012 15:09:30 +0100 CHAIR ... It should be able to handle any transformation of the concept, as in DRAW ME (or POINT TO/RECOGNIZE) A CHAIR IN TWO PIECES –.. ..SQUASHED ..IN PIECES -HALF VISIBLE ..WITH AN ARM MISSING ...WITH NO SEAT ..IN POLKA DOTS ...WITH RED STRIPES Concepts are designed for a world of everchanging, everevolving multiform objects (and actions). Semantic networks have zero creativity or adaptability – are applicable only to a uniform set of objects, (basically a database) - and also, crucially, have zero ability to physically recognize or interact with the relevant objects. I’ve been into it at length recently. You’re the one not paying attention. The suggestion that networks or similar can handle concepts is completely absurd. This is yet another form of the central problem of AGI, which you clearly do not understand – and I’m not trying to be abusive – I’ve been realising this again recently – people here are culturally punchdrunk with concepts like *concept* and *creativity*, and just don’t understand them in terms of AGI. From: Jim Bromer Sent: Tuesday, October 23, 2012 2:04 PM To: AGI Subject: Re: [agi] Re: Superficiality Produces Misunderstanding - Not Good Enough Mike Tintner <[email protected]> wrote: AI doesn’t handle concepts. Give me one example to prove that AI doesn't handle concepts. Jim Bromer On Tue, Oct 23, 2012 at 4:24 AM, Mike Tintner <[email protected]> wrote: Jim: Mike refuses to try to understand what I am saying because he would have to give up his sense of a superior point of view in order to understand it Concepts have nothing to do with semantic networks. AI doesn’t handle concepts. That is the challenge for AGI. The form of concepts is graphics. The referents of concepts are infinite realms.. What are you saying that is relevant to this, or that can challenge this – from any evidence? AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription 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-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
