Re:Fitting to Noise or Nothing at All: Machine Learning in M...
by mayank on Monday August 07, 2017 @01:03PM
(#14949072)
> Long story short, all "research" is pretty much B.S.
Easy there! When I start feeling cynical about things like almost every sentiment analysis paper, I go back and read some of my favorite papers, especially ones from the 70s and 80s (eg, [1])
Another useful heuristic is to read the best paper award winners from conferences in a particular area. It's not a perfect metric, but the signal/noise ratio is better.
Easy there! When I start feeling cynical about things like almost every sentiment analysis paper, I go back and read some of my favorite papers, especially ones from the 70s and 80s (eg, [1])
Another useful heuristic is to read the best paper award winners from conferences in a particular area. It's not a perfect metric, but the signal/noise ratio is better.
[1] the flajolet-martin algorithm: http://algo.inria.fr/flajolet/Publications/FlMa85.pdf