And error measures involving sum of absolute values (i.e., L1 norm) are central ...

kkylin on Nov 24, 2016 | parent | context | favorite | on: Why squared error? (2014)

And error measures involving sum of absolute values (i.e., L1 norm) are central to methods like lasso (https://en.wikipedia.org/wiki/Lasso_(statistics)) and their cousins.