I am pondering on the question in the title. As a human being, somehow I can sort a random sequence of numbers from 1 to $ 2^{2^{512}} $ in our universe in infinite time (But I am not sure.). Can an ML model do that in our universe if it is provided with infinite time? There is no restriction on how the learning algorithm is supposed to learn how to sort. (Be careful, even $ 2^{512} $ is bigger than the number of atoms in the universe. Therefore you will have limited memory.)

  • $\begingroup$ Are you basically asking whether a ML algorithm can learn to sort? What's the point of the range if you give the algorithm an infinite amount of time, can you elaborate a bit on that? $\endgroup$
    – SpiderRico
    Feb 6 at 3:11
  • $\begingroup$ @SpiderRico The point is that the number is very huge which cannot be represented by bits. I try to compare human cognition and ML models. It seems (not sure) that the computer cannot sort such big numbers because, after a point, the numbers cannot be represented even if you use all atoms in the universe to represent numbers. Can a human solve the problem without time limitation? $\endgroup$
    – verdery
    Feb 6 at 3:26
  • 2
    $\begingroup$ So you assume a memory limit then? Isn't then the question essentially boils down to whether we can store such a number in memory or not? I just can't see the ML part of the question here. $\endgroup$
    – SpiderRico
    Feb 6 at 3:55
  • $\begingroup$ The question is related to machine learning because I'm asking if an ML algorithm could theoretically learn how to sort any sequence of numbers (i.e. if it could learn some sorting algorithm). $\endgroup$
    – verdery
    Feb 6 at 4:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.