So, you minimize "total computational cost" by designing the whole machine solely to do bottom-up compression, with no pre-processed data, extraneous languages & processor architectures involved. http://www.cognitivealgorithm.info/2012/01/cognitive-algorithm.html
From: Ben Goertzel Sent: Wednesday, August 29, 2012 11:38 AM To: AGI Subject: Re: [agi] Finding the "Right" Computational Model to Support Occam's Razor Yes, you're right Abram, thanks... The question I raise is more interesting if one is looking at a "total computational cost" simplicity measure such as "the minimum amount of space and time resources needed to compute X" on a specific machine M.. I slightly edited my blog post to emphasize this... ben On Wed, Aug 29, 2012 at 2:46 AM, Abram Demski <[email protected]> wrote: Ben, If we're talking purely about description length, then I tend to side with Boris. Searching for the best model in a specific language is equivalent to searching for the best universal language: a Bayesian update of an Occam prior will keep the universality of the language (any possible next pattern is possible), but it will shift all the probabilities. So, a good algorithm will be effectively learning the optimal description language using lots of data. However, interestingly, the requirement that you give as an example is not a description-length requirement: What kind of criteria am I thinking of? I know I'm getting a little fuzzy at this point, but I'm thinking about stuff like: "Computing W and computing f(W) should take the same amount of space and time resources," for cases where it seems intuitively obvious that W and f(W) should take the same amount of space and time resources. This is a matter of efficiency, which is a completely different matter. There is no nice theorem saying that all universal Turing machines have the same time and space requirements for the same problems; in fact, they don't! However, there are some interesting theorems in this are which I've heard of... I don't know names, unfortunately. I've heard of theorems comparing the time and space requirements for universal Turing machines with different numbers of tapes, and also parallel computing with bounded numbers of processors (for example, cores polynomial in processing size). On Mon, Aug 27, 2012 at 8:34 AM, Ben Goertzel <[email protected]> wrote: Just some speculations about possible theoretical computer science I'd do if I had the time ;p http://multiverseaccordingtoben.blogspot.com/2012/08/finding-right-computational-model-to.html -- 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/303/7190161-766c6f07 Modify Your Subscription: https://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com -- Abram Demski http://lo-tho.blogspot.com/ AGI | Archives | Modify Your Subscription -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche 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
