eutectic

33 ms·

1. 301.▲
   by eutectic 10y ago

   If you have two predictors p0 and p1 then it can be shown that the mixture p(data) = (p0(data) + p1(data)) / 2 incurs at most 1 bit of loss compared to the better of p0 and p1. In other words, for any pair of machine learning algorithm
2. 302.▲
   by eutectic 10y ago

   The No Free lunch theorem is basically a consequence of the fact that almost all problems 'look random'; it doesn't really apply to the tiny subset that are of interest to humans.
3. 303.▲
   by eutectic 10y ago

   Aeroponics or hydroponics would help eliminate the mess issue.
4. 304.▲
   by eutectic 10y ago

   Or maybe it's a transcription error.
5. 305.▲
   by eutectic 10y ago

   Absolutely, I think that the failure to exploit spaced repetition is perhaps the most obvious and easily correctable flaw of most formal education. I strongly suspect that even just mixing questions from previous parts of the course into th
6. 306.▲
   by eutectic 10y ago

   What is your light exposure regimen? Even though I have a SAD lamp in my room set to turn on before my alarm I still struggle to regulate my sleep, especially in winter.
7. 307.▲
   by eutectic 10y ago

   I like to think about this algebraically: a1 = a0 ^ b0 b1 = b0 ^ a1 = b0 ^ (a0 ^ b0) = (a0 ^ b0) ^ b0 = a0 a2 = a1 ^ b1 = (a0 ^ b0) ^ a0 = (b0 ^ a0) ^ a0 = b0 The xor-linked list ( https://en.wikipedia.org/wi
8. 308.▲
   by eutectic 10y ago

   I think you mean a combination of sub and add (e.g. sub sub add). xor is somewhat special in that it is its own inverse.
9. 309.▲
   by eutectic 10y ago

   I mean you assign the value of the expression to your result. e.g: uint64_t x = 0; ... uint32_t y = (x+=x*x+9)>>32;
10. 310.▲
    by eutectic 10y ago

    Dieharder works OK, and you can easily install it on Ubuntu as a command-line tool. PractRand and TestU01 are more stringent and give you more knobs to play with.
11. 311.▲
    by eutectic 10y ago

    A 15 character 64 to 32-bit generator I devised which passes dieharder: (x+=x*x+9)>>32;
12. 312.▲
    by eutectic 10y ago

    *skin for sling. Oops.
13. 313.▲
    by eutectic 10y ago

    Is there a version of CFR for differentiable learners like neural networks?
14. 314.▲
    by eutectic 10y ago

    Using abrasives in acne products probably does more harm than good by irritating the sling and disrupting the acid mantle.
15. 315.▲
    by eutectic 10y ago

    Lozenges are also available. It's interesting to note that apparently nicotine is not addictive without the MOAIs from tobacco smoke.
16. 316.▲
    by eutectic 10y ago

    The density of gold at its melting point is actually 'only' 17.31 g/cm3.
17. 317.▲
    by eutectic 10y ago

    I think it would be too difficult/expensive to run enough trials to achieve statistical significance.
18. 318.▲
    by eutectic 10y ago

    I wonder how nicotine would fare.
19. 319.▲
    by eutectic 10y ago

    I wonder if it is possible to achieve super-human play purely using reinforcement learning to train deep neural networks.
20. 320.▲
    by eutectic 10y ago

    https://en.m.wikipedia.org/wiki/Tin_pest
21. 321.▲
    by eutectic 10y ago

    When my tolerance is low it doesn't take much to make me twitchy and impair my thinking.
22. 322.▲
    by eutectic 10y ago

    The charitable interpretation of his argument is that the outcome is only a statistical 'tie' if you accept the arbitrary significance threshold of p=5%. Even relatively weak evidence is still evidence. Unfortunately, some of the
23. 323.▲
    by eutectic 10y ago

    If you want people to be honest about their mistakes, they need to feel that doing so won't come back to bite them. The airline industry sets a great example in learning from failure to improve safety.
24. 324.▲
    by eutectic 10y ago

    This thread is literally the only Google search result for this phrase (for me)...
25. 325.▲
    by eutectic 10y ago

    Parrots can definitely understand words.
26. 326.▲
    by eutectic 10y ago

    Or: f <- [(+), (-)] let t = (xi `f` 7, xi `f` 11) (et, i) <- [t, swap t]
27. 327.▲
    by eutectic 10y ago

    I think this looks a lot nicer with Haskell-style do notation: import Control.Monad (guard) import Data.Ratio ((%)) results :: [[Int]] results = do th <- u o <- u r <- u guard
28. 328.▲
    by eutectic 10y ago

    Fool me once...
29. 329.▲
    by eutectic 10y ago

    There's nothing that makes typed languages inherently more verbose; a Haskell script will tend to be substantially shorter than the equivalent Python. Rust gives you much more control over how your program runs, but this necessarily me
30. 330.▲
    by eutectic 10y ago

    Presumably there is a pass to turn f(recur(x), y) into recur(f(y, acc), x) and then tail-call optimization can be applied. This works for any associative f.