News for hackers. Stuff that matters.
Re:Scientists narrow down the mechanism whereby cancer cells... by aceon48 on Tuesday October 17, 2017 @09:13AM (#15490603)
Ugh can we just put AlphaGo deep learning on diet / cancer/ food studies? One week meats are bad, next week meats are great dont eat carbs!

I'm getting javascript framework diet fad of the week fatigue

Re:Scientists narrow down the mechanism whereby cancer cells... by jmcqk6 on Tuesday October 17, 2017 @01:59PM (#15493106)
>One week meats are bad, next week meats are great dont eat carbs

I don't think this is a problem with our learning, but a problem with our unrelenting desire to compress complex subjects and findings into easily digestible morsels that people can understand without too much thought.

Deep learning isn't going to help with that.

Re:Scientists narrow down the mechanism whereby cancer cells... by pohl on Tuesday October 17, 2017 @09:45AM (#15490821)
Are we getting mixed signals about sugars, though? Proteins and carbs have had their ups & downs, but I don't recall ever hearing any praise for sugars.
Re:Scientists narrow down the mechanism whereby cancer cells... by hprotagonist on Tuesday October 17, 2017 @10:39AM (#15491264)
the problem with this approach is that the training data set just sucks.

humans are fully capable of doing the statistics to get proper answers on these questions without ML. we just cannot trust the data enough,because to get a reliable data set we would have to imprison a few tens of thousands of people for 60 years and control every aspect of their exercise and diet, and that sort of thing usually gets you arrested.