This has happened few times before. First comes some scientific understanding then it devolves into tinkering, "look what I got".
Renaming parts of machine learning as Computational learning theory was an attempt to distance the research from tinkering that produced AI winters. Leslie Valiant, Vladimir Vapnik and Alexey Chervonenkis produced new insight and new algorithms were produced.
Then Deep Learning stated to work again. Small scientific insights are shadowed by the massive amount of progress that comes with tinkering with parameters.
>For starters, he says, researchers should conduct "ablation studies" like those done with the translation algorithm: deleting parts of an algorithm one at a time to see the function of each component. He calls for "sliced analysis," in which an algorithm's performance is analyzed in detail to see how improvement in some areas might have a cost elsewhere.
...wait a minute. ML researchers don't already do this?!
Forgive me if I come across as arrogant or naive; I am a scientist but not in the ML field. I just can't fathom ANY field of research in which this very basic approach isn't followed.
3 comments
[ 3.6 ms ] story [ 27.6 ms ] threadRenaming parts of machine learning as Computational learning theory was an attempt to distance the research from tinkering that produced AI winters. Leslie Valiant, Vladimir Vapnik and Alexey Chervonenkis produced new insight and new algorithms were produced.
Then Deep Learning stated to work again. Small scientific insights are shadowed by the massive amount of progress that comes with tinkering with parameters.
...wait a minute. ML researchers don't already do this?!
Forgive me if I come across as arrogant or naive; I am a scientist but not in the ML field. I just can't fathom ANY field of research in which this very basic approach isn't followed.