Re: [agi] There is such a thing as a Free Lunch

I think we have to explore in order to learn - and in order to utilize our knowledge. But it is nicer when we can do so wisely.  So we explore areas that we do not know well in order to expand our knowledge which includes the application of knowledge that we have already acquired. But when we

Re: [agi] There is such a thing as a Free Lunch

On Saturday, October 03, 2020, at 10:11 AM, Jim Bromer wrote: > One issue that I have, is that you have to explore less likely possibilities > (when it is reasonably safe and reasonably ethical) in order to explore the > neighborhood. Hmm. Intelligence makes you explore fewer less likely paths,

Re: [agi] There is such a thing as a Free Lunch

So I guess my point of view would be that mathematics, in the contemporary common sense of the term, is not adequate for AGI. I do not think that human beings have ultimate compression-decompression systems that are responsible for thinking  although some kind of effective compression

Re: [agi] There is such a thing as a Free Lunch

I was thinking about it and the No Free Lunch theory is relevant to a large number of computational algorithms, but it is not a problem when the time differences (between runs of an algorithm) do not bother us. By thinking about simpler problems I was able to start to see some of the related

Re: [agi] There is such a thing as a Free Lunch

In regard to my post above, the question isn't if general intelligence exists, the real question is can you live forever? If not, then a general all around purpose solver doesn't surely exist -- Artificial General Intelligence List: AGI Permalink:

Re: [agi] There is such a thing as a Free Lunch

If you mean the ability to answer any question AND with perfect accuracy, the answer is no, you can't know everything because you'd need to know about the particles that make up "your know". It's a very different reality though than this, if we have a finite universe, then we can turn all the

Re: [agi] There is such a thing as a Free Lunch

I think the most relevant implication of the No Free Lunch Theorem here is that a general intelligence is mathematically impossible. Do you agree? On Thu, Oct 1, 2020 at 2:08 PM Jim Bromer wrote: > I am learning a little. The no free lunch theorem depends on the analysis > of an imperfect

Re: [agi] There is such a thing as a Free Lunch

I am learning a little. The no free lunch theorem depends on the analysis of an imperfect system that will lead to perfect knowledge.  But in real world it is possible that someone might come up with a solution to a complicated problem that would have to be heavily developed in order to be

Re: [agi] There is such a thing as a Free Lunch

On 9/30/20 6:21 PM, Jim Bromer wrote: Danko, I think your comment is closer to a reasonable definition but I am not sure that is the common definition.  The no free lunch idea is a very useful bit of insight but treating a rule that does not typically transcend the premises of its

Re: [agi] There is such a thing as a Free Lunch

Danko, I think your comment is closer to a reasonable definition but I am not sure that is the common definition.  The no free lunch idea is a very useful bit of insight but treating a rule that does not typically transcend the premises of its application as if it could is a mistake or at least

Re: [agi] There is such a thing as a Free Lunch

Hi Jim, I think that No Free Lunch is not about the impossibility to find new efficiencies. Rather, No free lunch is about what happens after this new efficiency has been implemented. Once you implement this new efficiency and once you can deal better with X, you have automatically limited

Re: [agi] There is such a thing as a Free Lunch

A 'free lunch' is possible because human beings do not know everything. That means that new efficiencies can be discovered.  Even though you may discuss a frame work as if it were the only basis to achieve a goal the argument does not prove the premise that the particular frame work you have in

Re: [agi] There is such a thing as a Free Lunch

Free lunch is a discount. Free lunch only happened when life stared on earth.  -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Ta433301e9ac5fb42-Mf96de8f142eba356523de092 Delivery options:

Re: [agi] There is such a thing as a Free Lunch

I guess Matt has a point, no predictor can predict everything, you "CAN" make a predictor that predicts the opposite of what it predicts (or just look to the fact that a machine can't know all of its particles details). Physics doesn't change of course and survival is progressed towards, and

Re: [agi] There is such a thing as a Free Lunch

" It's not because of the no free lunch theorem. It's because you can always find something that your program can't compress, and you have to add yet another special case. " I think it is exactly because of the No Free Lunch Theorem (NFLT). Let me explain how I think about it, and you can correct

Re: [agi] There is such a thing as a Free Lunch

Danko Nikolic says: "I see the no free lunch theorem striking every day. Every time we pick one ML architecture for one type of problem and another architecture for another type of problem, it is the No Free Lunch Theorem dictating the fact that we have to make thos chices and are not able to

Re: [agi] There is such a thing as a Free Lunch

It sounds like you are predicting that you will be unable to make any progress (in the versions of) ML (that you have in mind). -- Artificial General Intelligence List: AGI Permalink:

Re: [agi] There is such a thing as a Free Lunch

If there was no such thing as a free lunch all progress would have been impossible. By falling into your presumptions you lock yourself into them. -- Artificial General Intelligence List: AGI Permalink:

Re: [agi] There is such a thing as a Free Lunch

No Free Lunch only holds over finite sets. The incomputability of prediction holds over infinite sets. On Sun, Sep 27, 2020, 2:21 PM Danko Nikolic wrote: > But this is the No Free Lunch right there. > > On Sun, 27 Sep 2020, 20:05 Matt Mahoney wrote: > >> >> >> On Sun, Sep 27, 2020, 1:41 PM

Re: [agi] There is such a thing as a Free Lunch

But this is the No Free Lunch right there. On Sun, 27 Sep 2020, 20:05 Matt Mahoney wrote: > > > On Sun, Sep 27, 2020, 1:41 PM Danko Nikolic > wrote: > >> I see the no free lunch theorem striking every day. Every time we pick >> one ML architecture for one type of problem and another

Re: [agi] There is such a thing as a Free Lunch

On Sun, Sep 27, 2020, 1:41 PM Danko Nikolic wrote: > I see the no free lunch theorem striking every day. Every time we pick one > ML architecture for one type of problem and another architecture for > another type of problem, it is the No Free Lunch Theorem dictating the fact > that we have to

Re: [agi] There is such a thing as a Free Lunch

I see the no free lunch theorem striking every day. Every time we pick one ML architecture for one type of problem and another architecture for another type of problem, it is the No Free Lunch Theorem dictating the fact that we have to make thos chices and are not able to have one the same

Re: [agi] There is such a thing as a Free Lunch

On 2020-09-27 08:50:AM, Matt Mahoney wrote: On Sat, Sep 26, 2020, 10:32 PM TimTyler > wrote: On 2020-09-22 12:45:PM, Matt Mahoney wrote: > The no free lunch theorem is based on the false premise that it is > possible to have a uniform probability distribution

Re: [agi] There is such a thing as a Free Lunch

Dear all  I am writing to invite you to join the upcoming online Zoom seminars and research clinics:  7 Oct:  Prof Mark Fox, (Distinguished Professor of University of Toronto) Introduction to Knowledge Graphs, Ontologies and Ontology Engineering 15 Oct:  Prof John Macintyre, Pro Vice

Re: [agi] There is such a thing as a Free Lunch

Amen. I was definitely inspired by Hutter, but all my work utilized a stochastic, finite state model of computation. On Sun, 27 Sep 2020, mattmahone...@gmail.com wrote: On Sat, Sep 26, 2020, 10:32 PM TimTyler wrote: On 2020-09-22 12:45:PM, Matt Mahoney wrote: > The no free lunch

Re: [agi] There is such a thing as a Free Lunch

On Sat, Sep 26, 2020, 10:32 PM TimTyler wrote: > On 2020-09-22 12:45:PM, Matt Mahoney wrote: > > The no free lunch theorem is based on the false premise that it is > > possible to have a uniform probability distribution over an infinite > > set. The converse proves Occam's Razor. > > I don't

Re: [agi] There is such a thing as a Free Lunch

On 2020-09-22 12:45:PM, Matt Mahoney wrote: The no free lunch theorem is based on the false premise that it is possible to have a uniform probability distribution over an infinite set. The converse proves Occam's Razor. I don't think that's right. I looked here:

Re: [agi] There is such a thing as a Free Lunch

And it is relevant to the task and equipment that can be used on the task. My point is that there may be issues, like compression, that may be designed specifically for AI or AGI which may work better than more general compression methods. But that also means that they may not be as effective

Re: [agi] There is such a thing as a Free Lunch

It's basically just a meaningless saying in the end. Some things have cost, others don't. -- Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Ta433301e9ac5fb42-M8923d85a16a825ac1f4ae08a Delivery options:

Re: [agi] There is such a thing as a Free Lunch

On Tuesday, September 22, 2020, at 11:54 AM, Jim Bromer wrote: > Since you expend calories on things other than getting more calories then > that implies that every lunch that is sufficient to sustain you must be > partially free. You get more calories from lunch then you expend getting it. >

Re: [agi] There is such a thing as a Free Lunch

lcome.On Tuesday, September 22, 2020, at 10:37 AM, Jim Bromer wrote: > That is a very interesting point, but in talking about a uniform probability > distribution over an infinite set you are characterizing the issue as if it > were the universal characterization that underlies any

Re: [agi] There is such a thing as a Free Lunch

Since you expend calories on things other than getting more calories then that implies that every lunch that is sufficient to sustain you must be partially free. You get more calories from lunch then you expend getting it. Suppose that there was some supremely efficient characterization of a

Re: [agi] There is such a thing as a Free Lunch

That is a very interesting point, but in talking about a uniform probability distribution over an infinite set you are characterizing the issue as if it were the universal characterization that underlies any characterization (of all possible universal generalizations).

Re: [agi] There is such a thing as a Free Lunch

 Calories must be expended  to get more calories.  Calories must be expended to prevent the loss of calories.  Calories can be exchanged for any other unit of energy.  This axioms are pivotal in making a AGI energy management system.  No mater how elaborate a life form or conscious machine is it

Re: [agi] There is such a thing as a Free Lunch

The no free lunch theorem is based on the false premise that it is possible to have a uniform probability distribution over an infinite set. The converse proves Occam's Razor. On Tue, Sep 22, 2020, 11:39 AM Jim Bromer wrote: > If there was no such thing as a free lunch then we would all be

[agi] There is such a thing as a Free Lunch

If there was no such thing as a free lunch then we would all be living in the stone age. Every advancement is based on some kind of efficiency. Yes, those achievements come at a cost.  So there may be a relative trade-off but the loss of generality from a purely imaginary (unattainable ultimate