Alan Given your prednet example; In a 2-layer network, if the camera recording the R0 update frame was turned away for the update, the prediction made prior to the frame update would obviously be incorrect. Would the machine know - without referring to benchmarked footage - that the camera-input was moved to bias the result, and either report on it, or correct for it somehow, or just make a new prediction?
Rob ________________________________ From: Alan Grimes <[email protected]> Sent: Wednesday, 11 July 2018 6:03 PM To: Nanograte Knowledge Technologies via AGI Subject: Re: [agi] New Paper - Temporal Singularity and the Fermi Paradox Nanograte Knowledge Technologies via AGI wrote: > Look, I'm trying to follow the train of thought here with regards AGI. > It would be significant to AGI architecture if the contention was made > that first, existed consciousness, and then matter. It would probably > invert the life-cycle approaches we tend to follow in software > development. But all this informational noise is clouding the issue. > If it's not scientifically proven, no one can be right, or wrong. As > for consciousness, in the absence of a working architecture, most of > it is conjecture, and thus unprovable. However, there should be no > question that it slots into the top-5 AI questions. > > Rob BLEH!! OCCAM'S RAZOR, HACK CHOP, CUT SLICE DICE..... Proposal: Consciousness is precisely what is described by this: https://coxlab.github.io/prednet/ Task: "microgranualize" neural networks into cortical column-equivalent units then wire them up as described in the previous link -> collect a billion dollar prize. Altertaitively, to prove me wrong: Provide a simpler theory that explains more. or prove that this theory fails in some fundamental way. -- Please report bounces from this address to [email protected] Powers are not rights. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T81817474dba9a838-Macaf9844a8fb7930f2b26a54 Delivery options: https://agi.topicbox.com/groups
