News for hackers. Stuff that matters.

A comment on Google executives see cracks in their company’s success. Parent comment

Re:Google executives see cracks in their company’s success by selimthegrim on Monday June 21, 2021 @01:57PM (#27581739)
Where would you go now if you wanted to do such things? I have people in UW CS tell me a physics PhD in CMT is no proof I can learn how to analyze RNASeq data quickly enough for them and they want a postdoc with the experience already to teach DL to (I would have thought the other way around would be easier)
[ Parent | on HN ]
Re:Google executives see cracks in their company’s success by timy2shoes on Monday June 21, 2021 @02:34PM (#27582255)
The other way around is definitely easier. RNA-seq analysis is a mostly solved problem with DESeq2 (or edgeR/limma). The tutorials are very detailed. The most difficult part is experimental design, which you probably know already.

Deep learning, on the other hand, is so fraught with pitfalls and traps. Even if you can code up a model successfully, it's very easy to trick yourself that you're doing very well (see a previous discussion at https://news.ycombinator.com/item?id=27376839). In my opinion, most of the work should be spent on making sure that you're not tricking yourself.

[ Parent | on HN ]
Re:Google executives see cracks in their company’s success by dekhn on Monday June 21, 2021 @04:09PM (#27583736)
literally the first 5 years of training in any large scale data analysis should be "how not to trick yourself into thinking you found something significant that generalizes"
[ Parent | on HN ]
Re:Google executives see cracks in their company’s success by selimthegrim on Monday June 21, 2021 @06:46PM (#27585536)
So then why I am getting static from these PIs? Do they have some R01 report deadline a year away breathing down their necks?
[ Parent | on HN ]
Re:Google executives see cracks in their company’s success by dekhn on Monday June 21, 2021 @02:20PM (#27582034)
There's a reason I invented this saying: "The very best machine learning models in biology are distilled from postdoc tears".
[ Parent | on HN ]