1 paper · 1 filter
Marco Pollanen
Before a machine learning model ships, it often has to pass a suite of automated tests. Requiring every test to pass looks safe, yet it can reject many models that would have serve…