Benjamin Kuipers, a number of years ago, demonstrated a machine learning system that -- based on a 1D stream of perceptual symbols -- learned that this came from a 3D world. And this was a fairly general reinforcement-learning-based pattern learning system, not a highly specialized system oriented just toward learning dimensionality.... You can find the papers from this project on his website I'm sure...
The human brain appears not to have to carry out this particular act of learning, as knowledge of the dimensionality of space is wired into the brain in so many ways.. -- Ben g On Mon, Sep 10, 2012 at 3:08 PM, Anastasios Tsiolakidis <[email protected]> wrote: > All brain builders have to work with granularity in multiple ways, > including the question of how much "physics" should be inbuilt and how > much will have to be derived or discovered. I have previously asserted > that it is virtually a mathematical certainty that a "conditioned" > physical entity, one that is entangled with the physical world it > observes, cannot find out the true, deep laws of its universe, But > what about partial knowledge, such as the "3 dimensions". How can a > robot know in how many dimensions she is operating? Now, I am not > suggesting to including metrics from our mathematics like > http://www.wahl.org/fe/HTML_version/link/FE4W/c4.htm , even though > that would not be trivial at all either. What I am suggesting is that > the baby robot analyses inputs and outputs until it builds up a 3d > model of its world, as opposed to a 2d one or a 3.33 one. > > What could the preconditions be for the discovery of 3 dimensions? A > humanoid with many actuators and Degrees of Freedom would first of all > have to experiment with its own capabilities and multidimensional > "state" (know thyself) and use vision and interaction with the world > to "conclude" that it is in a 3d world - and the earth is flat! > Gravity will be tremendously helpful in such reasoning, but could the > discovery be also made in weightless space (or simulated weightless > space). With a few rockets and a pair of "eyes" like the Mars landers, > but only distant stars to look at, can you discover the 3 > dimensionality of the world? > > As I have said on plenty occasions, I expect such agnostic > intelligences to develop very alien concepts that will never, ever map > to "ours" and, lets say, the Turing test. But it would be nice to > know/guess how little/much data and stimuli a system/architecture > needs to start making sense of the world. It would be hard to suggest > that there is an animal on earth that does not have genetically > determined species-wide models of the world, though I'd wager dogs and > humans operate with a 2d map, not unlike Dogville! Which again is very > different both from knowing that the world is 2d and from discovering > the world's two dimensions. > > AT > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/212726-11ac2389 > Modify Your Subscription: https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche ------------------------------------------- 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
