With Yann on most things here, except for “Learning algorithms themselves are not biased”. That’s wrong. The choice of network structure you use, the regularization you apply, for eg, are all introducing biases from from certain kinds of decisions never being made.https://twitter.com/ylecun/status/1203211859366576128 …
-
-
Too much people seems to believe that the NFL theorem will protect their algorithm from generalizing poorly. But if studying loss landscapes have taught me something, it's that different algorithms have different heuristic bias.
-
This is cool, because one can use this, but it's also the possibility of a bias. However, I'm not sure that the possibility that it translates to a social bias at the end comes with an high probability.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.