• jeffhykin@lemm.ee
    link
    fedilink
    English
    arrow-up
    3
    arrow-down
    1
    ·
    edit-2
    8 months ago

    This is why my field (reinforcement learning) is unfortunately not science.

    (Can’t really publish “hey I tried this algorithm and it didn’t work”)

      • jeffhykin@lemm.ee
        link
        fedilink
        English
        arrow-up
        2
        ·
        edit-2
        8 months ago

        I guess I should’ve clarified; in reforcement learning “I was wrong in numerous ways” almost always translates to “unpublishable, try to not be wrong next time”. Nobody cares if a reinforcement learning hypothesis didn’t work, its only worth publishing if it worked well.

        • overcast5348@lemmy.world
          link
          fedilink
          English
          arrow-up
          3
          ·
          8 months ago

          Gotcha.

          I thought that was the norm in all academia these days? Can a physicist (or anyone from another field) publish results that didn’t go as expected and save future scientists some time?