Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

On Mon, Jun 29, 2020 at 9:43 AM Matt Mahoney wrote: > The problem with Occam's Razor or algorithmic information theory is > that simplicity is language dependent. In practice, that's for time complexity _only_. In practice, the data founding our models are vastly in excess of the size of the s

Re: [agi] Re: Minsky's Physics Envy

Depending on what you mean by a soul, it either has an effect on human behavior or it doesn't. If it does, then you can model this effect in software and create machines with souls. If it doesn't, then souls are irrelevant to AGI. But maybe you want to know if I create a robot that looks and acts

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

Carefully crafted masterpieces of strings also have an aesthetic effect. Like the one Matt created. You might not know immediately what they mean... but then it sinks in. It's art too. Some strings you never know what they mean but the aesthetic effect impresses. For example some crazy mathemati

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

No, the simple trick just means to remove things - things that don't affect the result enough to kill you. I give you 2 solutions to buying candy: 1) Hand cash to the candy man and take the candy and eat it. 2) Hand cash to the candy man and take the candy then place it on the floor then pick it

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

"As simple as possible but no simpler" (I believe Einstein's paraphrase of Occam's Razor), means not so simple that the theory disagrees with observation. Occam's Razor is true because any possible probability distribution over a set of strings (descriptions, theories, programs) must favor shorter

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

Still though you have to have a computational resource topology to process the theories. We immediately assume a single executable program funneled down to CPU registers as Turing machine emulation but as there's more strings and theories the resources need to distribute on the computational com

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

You right! Thanks for reposting that Matt quote I somehow missed it so I reread a few times - he actually pre-answered the question I asked afterwards, uh duh.  -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T377

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

I hear someone in the background screaming "simpl please" More people can read it if it's simple and unified, come on guys! I already believe I fully understand all you said Matt, it's just "simple but not too simple". And that you start at Brute Force and use hints where to search (as

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

> Occam's Razor is true because for all possible probability distributions over > the infinite set of possible theories described by strings, each theory can > only be more likely than a finite set of longer theories. This is true in any > language used to describe the theories. You completely

Re: [agi] Re: I made a multi-file compressor that beats 7zip on real-world data

@James Ooooh... the Sequitur algorithm , developed in 1997:* "*a recursive algorithm developed by Craig Nevill-Manning  and Ian H. Witten  in 1997 t

Re: [agi] Re: I made a multi-file compressor that beats 7zip on real-world data

Regarding the algorithm itself, I am now using it on a byte level instead of lines level, so it works on any kind of data. The algorithm's runtime is indeed linear in the input length, which still amazes me. How is compression in linear time even possible? Yet it is. The basic trick is that by

Re: [agi] Re: I made a multi-file compressor that beats 7zip on real-world data

@Matt Yes, possibly similar. What I do is probably even simpler as all my productions are just pairs of symbols (each being either a terminals or a non-terminal). -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

Matt, I know that, the answer is more likely the simplest explanation one, 2 4 8 16 32.that's what I said above. So is my use of the word razor correct then? - Simple but not too simple. -- Artificial General Intelligence List: AGI Permalink: https://a

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

On Tuesday, June 30, 2020, at 8:38 AM, Matt Mahoney wrote: > You could write a program to print any next number you wanted. But both the > code and the English language description would be longer. Occam's Razor and > Solomonoff induction says the answer is 32. That's an absolute answer but vari

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

What is the next number in the sequence 1 2 4 8 16? The answer is 30, the number of positive integers that are factors of n! Or is it 31, the number of regions of 4-space divided by n hyperplanes? Or is it 32, the next number printed by: for(int i=0;;i*=2)cout

Re: [agi] Re: Boogaloo news.

This is AGI related. Judging by the way China handled the accident at the Wuhan lab I don't know if I trust China  handling any sort of existential AI risk or grey goo accident. This topic used to be often discussed on this forum but not so much now... --

Re: [agi] Re: Goertzel's "Grounding Occam's Razor in a Formal Theory of Simplicity"

On short and accurately simulable, or computable in reality: https://www.youtube.com/watch?v=MMiKyfd6hA0 -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T37756381803ac879-M3deea99027c5928d5ae87e68 Delivery options